You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@jclouds.apache.org by Battula Kishore <ba...@adobe.com.INVALID> on 2017/06/06 03:42:51 UTC
Re: Adding async support for Jclouds library
Hi Andrew,
Any update on this?
-- Thanks
-- Kishore
On 29/05/17, 1:06 PM, "Battula Kishore" <ba...@adobe.com.INVALID> wrote:
>Hi Andrew,
>
>Thanks andrew for the quick response. Here is the GitHub repo and instructions on how to run the tests https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fkishore25kumar%2Fs3proxy-async-test-setup&data=02%7C01%7C%7C8969ed65c6a749aebfa608d4a66583cc%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636316402308126519&sdata=0YESeCSlmChUOicpe2m9SYC1WaB7m%2B7hIB6o%2BTOp1Jo%3D&reserved=0. The READE.md also has the repo details of s3proxy as well as jclouds implementation. Hope this helps? Let me know if you need anything else?
>
>In the mean time you review these results if you can let me know the design review process I can be prepared for that.
>
>-- Thanks
>-- Kishore
>
>
>
>
>
>
>
>
>
>On 26/05/17, 12:40 PM, "Andrew Gaul" <ga...@apache.org> wrote:
>
>>Kishore, these are promising results! I reformatted the most important
>>rows which show a 2x improvement in throughput and latency:
>>
>>10 10,000 Async Http Lib 209 282 48
>>10 10,000 OutputStream 392 542 25
>>
>>Can you share the implementation and include instructions on how to
>>replicate these tests?
>>
>>On Wed, May 24, 2017 at 05:50:52AM +0000, Battula Kishore wrote:
>>> Hi,
>>>
>>> This is Kishore who is working on async poc using mail id(kishore25kumar@gmail.com<ma...@gmail.com>). I work at adobe and we wanted to implement async support for jclouds library and contribute it back.
>>>
>>> From the last discussion I was asked to get the performance numbers for the two approaches.
>>> Approach 1: Using Http Async Library
>>> Approach 2: Using Outputstream
>>>
>>> Test setup:
>>>
>>> 1. Both the s3 proxy server and test runner are running in same Docker container in azure west-us region.
>>>
>>> 2. Azure storage account is also residing in same west-us region.
>>>
>>> 3. A bucket is prepopulated with 100,000 files, each file of 1 MB size before test start.
>>>
>>> 4. The test runner sends unique requests to s3proxy to download files.
>>>
>>> Virtual Machine spec: CPU - 8 cores, Memory - 28 GB (Standard_D4 Azure machine)
>>>
>>> S3proxy is running with 1 jetty worker thread in all the scenarios. The payload size used is 1 MB file. Here are the performance numbers.
>>> Test Runner Threads
>>>
>>> Iteration Per thread
>>>
>>> Approach
>>>
>>> Avg response time (ms)
>>>
>>> 99%tile time (ms)
>>>
>>> Throughput
>>> (Requests / sec)
>>>
>>> 1
>>>
>>> 10,000
>>>
>>> Async Http Lib
>>>
>>> 45
>>>
>>> 87
>>>
>>> 22
>>>
>>> 5
>>>
>>> 10,000
>>>
>>> Async Http Lib
>>>
>>> 107
>>>
>>> 159
>>>
>>> 47
>>>
>>> 10
>>>
>>> 10,000
>>>
>>> Async Http Lib
>>>
>>> 209
>>>
>>> 282
>>>
>>> 48
>>>
>>> 1
>>>
>>> 10,000
>>>
>>> OutputStream
>>>
>>> 41
>>>
>>> 85
>>>
>>> 24
>>>
>>> 5
>>>
>>> 10,000
>>>
>>> OutputStream
>>>
>>> 190
>>>
>>> 283
>>>
>>> 26
>>>
>>> 10
>>>
>>> 10,000
>>>
>>> OutputStream
>>>
>>> 392
>>>
>>> 542
>>>
>>> 25
>>>
>>>
>>> Summary: Under load Http Async Library approach is providing more throughput compared to Output stream approach.
>>>
>>>
>>> Both the approaches improve performance. The output stream approach can be used along with Http Async library approach which is giving around (3-5 ms) improvement in latency.
>>>
>>> Each approach is independent development. At this point I am keen to take up Http Async Library development.
>>>
>>>
>>> -- Thanks
>>> -- Kishore
>>>
>>
>>--
>>Andrew Gaul
>>https://na01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fgaul.org%2F&data=02%7C01%7C%7C8690ee3e00bc4e3da0e608d4a406565e%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636313794511467457&sdata=XQ%2FshVjdqC3KiVEuyH6%2FJvmDN5DHBmS0kIBx98V89KY%3D&reserved=0
Re: Adding async support for Jclouds library
Posted by Battula Kishore <ba...@adobe.com.INVALID>.
Hi Ignasi,
I have raised a pull request for async put blob api in Jclouds git. For reference https://github.com/jclouds/jclouds/pull/1114
-- Thanks
-- Kishore
On 15/06/17, 10:53 AM, "Battula Kishore" <ba...@adobe.com.INVALID> wrote:
>Hi Ignasi,
>Thanks. I will cleanup and will raise a pull request directly to Jclouds in couple of days.
>
>-- Thanks
>-- Kishore
>
>
>
>
>
>
>
>
>
>
>On 14/06/17, 3:19 PM, "Ignasi Barrera" <na...@apache.org> wrote:
>
>>I participated in both reviews and I'm happy with the status of the
>>code in the PRs. I'd say you clean things up and open directly a PR to
>>the jclouds repo. If there is anything to discuss further, we can do
>>it there. As mentioned else-thread, the implementation should add
>>async support for all HTTP drivers that support it, so I'd like to see
>>the changes tot he OkHttp driver too in the PR.
>>
>>Keep up your fantastic job!
>>
>>
>>I.
>>
>>On 13 June 2017 at 11:27, Battula Kishore <ba...@adobe.com.invalid> wrote:
>>> Hi Ignasi,
>>>
>>> Just to do a quick walkthrough what I have already did towards the async approach.
>>> Here is the initial pull request for AsyncPOC
>>> https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2FSpandanThakur%2Fjclouds%2Fpull%2F2&data=02%7C01%7C%7Cc47d52d09b5846a8a9e008d4b30aa82e%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636330305747272162&sdata=Z0JPWGfCL9m8JQpYGlXQnJNp2WHJeWPBJp4WCf4fHVs%3D&reserved=0
>>> This pull request has gone through couple of reviews from the community members and I have made necessary changes. One last thing I was asked is to implement AsyncBlobInterface instead of adding it to existing BlobInterface. I have done it in this pull request.
>>> https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2FSpandanThakur%2Fjclouds%2Fpull%2F4&data=02%7C01%7C%7Cc47d52d09b5846a8a9e008d4b30aa82e%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636330305747272162&sdata=tkq5epIrtaoLdILEV6A3u8q%2BiZpx0wSfenzZFMY%2BO4Y%3D&reserved=0
>>>
>>> I am looking for some inputs on how to proceed from here to get it into a state where it can be merged. Then after the I can follow the same steps for other API's as well once I get familiar with the process for one API. Looking forward for your inputs.
>>>
>>> --Thanks
>>> --Kishore
>>>
>>> -----Original Message-----
>>> From: Ignasi Barrera [mailto:nacx@apache.org]
>>> Sent: Monday, June 12, 2017 11:47 PM
>>> To: dev@jclouds.apache.org
>>> Subject: Re: Adding async support for Jclouds library
>>>
>>> Hi!
>>>
>>> I think that given the output from your benchmarks it is OK to proceed with the async approach. The asnyc and OutputStream versions are not exclusive and the benchmarks show no reason for not starting with implementation of the async feature.
>>>
>>> For me, let's go with the next step and open a pull request with the PoC code/continue the discussion/whatever plan you had in mind :)
>>>
>>>
>>> I.
>>>
>>> On Jun 12, 2017 4:48 AM, "Battula Kishore" <ba...@adobe.com.invalid>
>>> wrote:
>>>
>>> Hi Andrew, Ignasi
>>>
>>> Checking once again any update on this?
>>>
>>> -- Thanks
>>> -- Kishore
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>>
>>> On 06/06/17, 12:10 PM, "Ignasi Barrera" <na...@apache.org> wrote:
>>>
>>>>Hi,
>>>>
>>>>I can pick it up, but I think we should have some more feedback from
>>>>Andrew first, at least to see if he still has some concerns about the
>>>>implementation.
>>>>I do like the current async approach but it wouldn't be right to move
>>>>forward without the OK of the team member that had concerns and has
>>>>been more involved in the design discussion.
>>>>
>>>>
>>>>I,
>>>>
>>>>On 6 June 2017 at 08:28, Felix Meschberger <fm...@adobe.com.invalid>
>>> wrote:
>>>>> Hi Jclouds dev
>>>>>
>>>>> Anyone else who could pick up instead of AndrewG ?
>>>>>
>>>>> It would be great to be able to have this support added to JClouds.
>>>>>
>>>>> Thanks
>>>>> Felix
>>>>>
>>>>>> Am 06.06.2017 um 05:49 schrieb Andrew Phillips <an...@apache.org>:
>>>>>>
>>>>>> Hi Kishore
>>>>>>
>>>>>> Andrew G. is travelling for most of June, as far as I understand, so
>>> will likely be a bit slower to respond. Thanks for your patience!
>>>>>>
>>>>>> ap
>>>>>>
>>>>>> On 2017-06-05 23:42, Battula Kishore wrote:
>>>>>>> Hi Andrew,
>>>>>>> Any update on this?
>>>>>>> -- Thanks
>>>>>>> -- Kishore
>>>>>>> On 29/05/17, 1:06 PM, "Battula Kishore" <ba...@adobe.com.INVALID>
>>> wrote:
>>>>>>>> Hi Andrew,
>>>>>>>> Thanks andrew for the quick response. Here is the GitHub repo and
>>> instructions on how to run the tests https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fna01.safelinks&data=02%7C01%7C%7C6ad1a11f02da4249a9df08d4b1bf4312%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636328882406496498&sdata=VxmspqySHr7onsxPr75PVxGzQgfEVVJoITrOfy5BdHU%3D&reserved=0.
>>> protection.outlook.com/?url=https%3A%2F%2Fgithub.com%
>>> 2Fkishore25kumar%2Fs3proxy-async-test-setup&data=02%7C01%7C%
>>> 7C8969ed65c6a749aebfa608d4a66583cc%7Cfa7b1b5a7b34438794aed2c178de
>>> cee1%7C0%7C0%7C636316402308126519&sdata=0YESeCSlmChUOicpe2m9SYC1WaB7m%
>>> 2B7hIB6o%2BTOp1Jo%3D&reserved=0. The READE.md also has the repo details of s3proxy as well as jclouds implementation. Hope this helps? Let me know if you need anything else?
>>>>>>>> In the mean time you review these results if you can let me know
>>>>>>>> the
>>> design review process I can be prepared for that.
>>>>>>>> -- Thanks
>>>>>>>> -- Kishore
>>>>>>>> On 26/05/17, 12:40 PM, "Andrew Gaul" <ga...@apache.org> wrote:
>>>>>>>>> Kishore, these are promising results! I reformatted the most
>>> important
>>>>>>>>> rows which show a 2x improvement in throughput and latency:
>>>>>>>>> 10 10,000 Async Http Lib 209 282 48
>>>>>>>>> 10 10,000 OutputStream 392 542 25
>>>>>>>>> Can you share the implementation and include instructions on how
>>>>>>>>> to replicate these tests?
>>>>>>>>> On Wed, May 24, 2017 at 05:50:52AM +0000, Battula Kishore wrote:
>>>>>>>>>> Hi,
>>>>>>>>>> This is Kishore who is working on async poc using mail id(
>>> kishore25kumar@gmail.com<ma...@gmail.com>). I work at adobe and we wanted to implement async support for jclouds library and contribute it back.
>>>>>>>>>> From the last discussion I was asked to get the performance
>>>>>>>>>> numbers
>>> for the two approaches.
>>>>>>>>>> Approach 1: Using Http Async Library Approach 2: Using
>>>>>>>>>> Outputstream Test setup:
>>>>>>>>>> 1. Both the s3 proxy server and test runner are running in
>>> same Docker container in azure west-us region.
>>>>>>>>>> 2. Azure storage account is also residing in same west-us
>>> region.
>>>>>>>>>> 3. A bucket is prepopulated with 100,000 files, each file of
>>> 1 MB size before test start.
>>>>>>>>>> 4. The test runner sends unique requests to s3proxy to
>>> download files.
>>>>>>>>>> Virtual Machine spec: CPU - 8 cores, Memory - 28 GB (Standard_D4
>>> Azure machine)
>>>>>>>>>> S3proxy is running with 1 jetty worker thread in all the scenarios.
>>> The payload size used is 1 MB file. Here are the performance numbers.
>>>>>>>>>> Test Runner Threads
>>>>>>>>>> Iteration Per thread
>>>>>>>>>> Approach
>>>>>>>>>> Avg response time (ms)
>>>>>>>>>> 99%tile time (ms)
>>>>>>>>>> Throughput
>>>>>>>>>> (Requests / sec)
>>>>>>>>>> 1
>>>>>>>>>> 10,000
>>>>>>>>>> Async Http Lib
>>>>>>>>>> 45
>>>>>>>>>> 87
>>>>>>>>>> 22
>>>>>>>>>> 5
>>>>>>>>>> 10,000
>>>>>>>>>> Async Http Lib
>>>>>>>>>> 107
>>>>>>>>>> 159
>>>>>>>>>> 47
>>>>>>>>>> 10
>>>>>>>>>> 10,000
>>>>>>>>>> Async Http Lib
>>>>>>>>>> 209
>>>>>>>>>> 282
>>>>>>>>>> 48
>>>>>>>>>> 1
>>>>>>>>>> 10,000
>>>>>>>>>> OutputStream
>>>>>>>>>> 41
>>>>>>>>>> 85
>>>>>>>>>> 24
>>>>>>>>>> 5
>>>>>>>>>> 10,000
>>>>>>>>>> OutputStream
>>>>>>>>>> 190
>>>>>>>>>> 283
>>>>>>>>>> 26
>>>>>>>>>> 10
>>>>>>>>>> 10,000
>>>>>>>>>> OutputStream
>>>>>>>>>> 392
>>>>>>>>>> 542
>>>>>>>>>> 25
>>>>>>>>>> Summary: Under load Http Async Library approach is providing
>>>>>>>>>> more
>>> throughput compared to Output stream approach.
>>>>>>>>>> Both the approaches improve performance. The output stream
>>>>>>>>>> approach
>>> can be used along with Http Async library approach which is giving around
>>> (3-5 ms) improvement in latency.
>>>>>>>>>> Each approach is independent development. At this point I am
>>>>>>>>>> keen
>>> to take up Http Async Library development.
>>>>>>>>>> -- Thanks
>>>>>>>>>> -- Kishore
>>>>>>>>> --
>>>>>>>>> Andrew Gaul
>>>>>>>>> https://na01.safelinks.protection.outlook.com/?url=
>>> http%3A%2F%2Fgaul.org%2F&data=02%7C01%7C%7C8690ee3e00bc4e3da0e608d4a406565e%
>>> 7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636313794511467457&sdata=XQ%
>>> 2FshVjdqC3KiVEuyH6%2FJvmDN5DHBmS0kIBx98V89KY%3D&reserved=0
>>>>>
Re: Adding async support for Jclouds library
Posted by Battula Kishore <ba...@adobe.com.INVALID>.
Hi Ignasi,
Thanks. I will cleanup and will raise a pull request directly to Jclouds in couple of days.
-- Thanks
-- Kishore
On 14/06/17, 3:19 PM, "Ignasi Barrera" <na...@apache.org> wrote:
>I participated in both reviews and I'm happy with the status of the
>code in the PRs. I'd say you clean things up and open directly a PR to
>the jclouds repo. If there is anything to discuss further, we can do
>it there. As mentioned else-thread, the implementation should add
>async support for all HTTP drivers that support it, so I'd like to see
>the changes tot he OkHttp driver too in the PR.
>
>Keep up your fantastic job!
>
>
>I.
>
>On 13 June 2017 at 11:27, Battula Kishore <ba...@adobe.com.invalid> wrote:
>> Hi Ignasi,
>>
>> Just to do a quick walkthrough what I have already did towards the async approach.
>> Here is the initial pull request for AsyncPOC
>> https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2FSpandanThakur%2Fjclouds%2Fpull%2F2&data=02%7C01%7C%7Cc47d52d09b5846a8a9e008d4b30aa82e%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636330305747272162&sdata=Z0JPWGfCL9m8JQpYGlXQnJNp2WHJeWPBJp4WCf4fHVs%3D&reserved=0
>> This pull request has gone through couple of reviews from the community members and I have made necessary changes. One last thing I was asked is to implement AsyncBlobInterface instead of adding it to existing BlobInterface. I have done it in this pull request.
>> https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2FSpandanThakur%2Fjclouds%2Fpull%2F4&data=02%7C01%7C%7Cc47d52d09b5846a8a9e008d4b30aa82e%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636330305747272162&sdata=tkq5epIrtaoLdILEV6A3u8q%2BiZpx0wSfenzZFMY%2BO4Y%3D&reserved=0
>>
>> I am looking for some inputs on how to proceed from here to get it into a state where it can be merged. Then after the I can follow the same steps for other API's as well once I get familiar with the process for one API. Looking forward for your inputs.
>>
>> --Thanks
>> --Kishore
>>
>> -----Original Message-----
>> From: Ignasi Barrera [mailto:nacx@apache.org]
>> Sent: Monday, June 12, 2017 11:47 PM
>> To: dev@jclouds.apache.org
>> Subject: Re: Adding async support for Jclouds library
>>
>> Hi!
>>
>> I think that given the output from your benchmarks it is OK to proceed with the async approach. The asnyc and OutputStream versions are not exclusive and the benchmarks show no reason for not starting with implementation of the async feature.
>>
>> For me, let's go with the next step and open a pull request with the PoC code/continue the discussion/whatever plan you had in mind :)
>>
>>
>> I.
>>
>> On Jun 12, 2017 4:48 AM, "Battula Kishore" <ba...@adobe.com.invalid>
>> wrote:
>>
>> Hi Andrew, Ignasi
>>
>> Checking once again any update on this?
>>
>> -- Thanks
>> -- Kishore
>>
>>
>>
>>
>>
>>
>>
>>
>>
>> On 06/06/17, 12:10 PM, "Ignasi Barrera" <na...@apache.org> wrote:
>>
>>>Hi,
>>>
>>>I can pick it up, but I think we should have some more feedback from
>>>Andrew first, at least to see if he still has some concerns about the
>>>implementation.
>>>I do like the current async approach but it wouldn't be right to move
>>>forward without the OK of the team member that had concerns and has
>>>been more involved in the design discussion.
>>>
>>>
>>>I,
>>>
>>>On 6 June 2017 at 08:28, Felix Meschberger <fm...@adobe.com.invalid>
>> wrote:
>>>> Hi Jclouds dev
>>>>
>>>> Anyone else who could pick up instead of AndrewG ?
>>>>
>>>> It would be great to be able to have this support added to JClouds.
>>>>
>>>> Thanks
>>>> Felix
>>>>
>>>>> Am 06.06.2017 um 05:49 schrieb Andrew Phillips <an...@apache.org>:
>>>>>
>>>>> Hi Kishore
>>>>>
>>>>> Andrew G. is travelling for most of June, as far as I understand, so
>> will likely be a bit slower to respond. Thanks for your patience!
>>>>>
>>>>> ap
>>>>>
>>>>> On 2017-06-05 23:42, Battula Kishore wrote:
>>>>>> Hi Andrew,
>>>>>> Any update on this?
>>>>>> -- Thanks
>>>>>> -- Kishore
>>>>>> On 29/05/17, 1:06 PM, "Battula Kishore" <ba...@adobe.com.INVALID>
>> wrote:
>>>>>>> Hi Andrew,
>>>>>>> Thanks andrew for the quick response. Here is the GitHub repo and
>> instructions on how to run the tests https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fna01.safelinks&data=02%7C01%7C%7C6ad1a11f02da4249a9df08d4b1bf4312%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636328882406496498&sdata=VxmspqySHr7onsxPr75PVxGzQgfEVVJoITrOfy5BdHU%3D&reserved=0.
>> protection.outlook.com/?url=https%3A%2F%2Fgithub.com%
>> 2Fkishore25kumar%2Fs3proxy-async-test-setup&data=02%7C01%7C%
>> 7C8969ed65c6a749aebfa608d4a66583cc%7Cfa7b1b5a7b34438794aed2c178de
>> cee1%7C0%7C0%7C636316402308126519&sdata=0YESeCSlmChUOicpe2m9SYC1WaB7m%
>> 2B7hIB6o%2BTOp1Jo%3D&reserved=0. The READE.md also has the repo details of s3proxy as well as jclouds implementation. Hope this helps? Let me know if you need anything else?
>>>>>>> In the mean time you review these results if you can let me know
>>>>>>> the
>> design review process I can be prepared for that.
>>>>>>> -- Thanks
>>>>>>> -- Kishore
>>>>>>> On 26/05/17, 12:40 PM, "Andrew Gaul" <ga...@apache.org> wrote:
>>>>>>>> Kishore, these are promising results! I reformatted the most
>> important
>>>>>>>> rows which show a 2x improvement in throughput and latency:
>>>>>>>> 10 10,000 Async Http Lib 209 282 48
>>>>>>>> 10 10,000 OutputStream 392 542 25
>>>>>>>> Can you share the implementation and include instructions on how
>>>>>>>> to replicate these tests?
>>>>>>>> On Wed, May 24, 2017 at 05:50:52AM +0000, Battula Kishore wrote:
>>>>>>>>> Hi,
>>>>>>>>> This is Kishore who is working on async poc using mail id(
>> kishore25kumar@gmail.com<ma...@gmail.com>). I work at adobe and we wanted to implement async support for jclouds library and contribute it back.
>>>>>>>>> From the last discussion I was asked to get the performance
>>>>>>>>> numbers
>> for the two approaches.
>>>>>>>>> Approach 1: Using Http Async Library Approach 2: Using
>>>>>>>>> Outputstream Test setup:
>>>>>>>>> 1. Both the s3 proxy server and test runner are running in
>> same Docker container in azure west-us region.
>>>>>>>>> 2. Azure storage account is also residing in same west-us
>> region.
>>>>>>>>> 3. A bucket is prepopulated with 100,000 files, each file of
>> 1 MB size before test start.
>>>>>>>>> 4. The test runner sends unique requests to s3proxy to
>> download files.
>>>>>>>>> Virtual Machine spec: CPU - 8 cores, Memory - 28 GB (Standard_D4
>> Azure machine)
>>>>>>>>> S3proxy is running with 1 jetty worker thread in all the scenarios.
>> The payload size used is 1 MB file. Here are the performance numbers.
>>>>>>>>> Test Runner Threads
>>>>>>>>> Iteration Per thread
>>>>>>>>> Approach
>>>>>>>>> Avg response time (ms)
>>>>>>>>> 99%tile time (ms)
>>>>>>>>> Throughput
>>>>>>>>> (Requests / sec)
>>>>>>>>> 1
>>>>>>>>> 10,000
>>>>>>>>> Async Http Lib
>>>>>>>>> 45
>>>>>>>>> 87
>>>>>>>>> 22
>>>>>>>>> 5
>>>>>>>>> 10,000
>>>>>>>>> Async Http Lib
>>>>>>>>> 107
>>>>>>>>> 159
>>>>>>>>> 47
>>>>>>>>> 10
>>>>>>>>> 10,000
>>>>>>>>> Async Http Lib
>>>>>>>>> 209
>>>>>>>>> 282
>>>>>>>>> 48
>>>>>>>>> 1
>>>>>>>>> 10,000
>>>>>>>>> OutputStream
>>>>>>>>> 41
>>>>>>>>> 85
>>>>>>>>> 24
>>>>>>>>> 5
>>>>>>>>> 10,000
>>>>>>>>> OutputStream
>>>>>>>>> 190
>>>>>>>>> 283
>>>>>>>>> 26
>>>>>>>>> 10
>>>>>>>>> 10,000
>>>>>>>>> OutputStream
>>>>>>>>> 392
>>>>>>>>> 542
>>>>>>>>> 25
>>>>>>>>> Summary: Under load Http Async Library approach is providing
>>>>>>>>> more
>> throughput compared to Output stream approach.
>>>>>>>>> Both the approaches improve performance. The output stream
>>>>>>>>> approach
>> can be used along with Http Async library approach which is giving around
>> (3-5 ms) improvement in latency.
>>>>>>>>> Each approach is independent development. At this point I am
>>>>>>>>> keen
>> to take up Http Async Library development.
>>>>>>>>> -- Thanks
>>>>>>>>> -- Kishore
>>>>>>>> --
>>>>>>>> Andrew Gaul
>>>>>>>> https://na01.safelinks.protection.outlook.com/?url=
>> http%3A%2F%2Fgaul.org%2F&data=02%7C01%7C%7C8690ee3e00bc4e3da0e608d4a406565e%
>> 7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636313794511467457&sdata=XQ%
>> 2FshVjdqC3KiVEuyH6%2FJvmDN5DHBmS0kIBx98V89KY%3D&reserved=0
>>>>
Re: Adding async support for Jclouds library
Posted by Ignasi Barrera <na...@apache.org>.
I participated in both reviews and I'm happy with the status of the
code in the PRs. I'd say you clean things up and open directly a PR to
the jclouds repo. If there is anything to discuss further, we can do
it there. As mentioned else-thread, the implementation should add
async support for all HTTP drivers that support it, so I'd like to see
the changes tot he OkHttp driver too in the PR.
Keep up your fantastic job!
I.
On 13 June 2017 at 11:27, Battula Kishore <ba...@adobe.com.invalid> wrote:
> Hi Ignasi,
>
> Just to do a quick walkthrough what I have already did towards the async approach.
> Here is the initial pull request for AsyncPOC
> https://github.com/SpandanThakur/jclouds/pull/2
> This pull request has gone through couple of reviews from the community members and I have made necessary changes. One last thing I was asked is to implement AsyncBlobInterface instead of adding it to existing BlobInterface. I have done it in this pull request.
> https://github.com/SpandanThakur/jclouds/pull/4
>
> I am looking for some inputs on how to proceed from here to get it into a state where it can be merged. Then after the I can follow the same steps for other API's as well once I get familiar with the process for one API. Looking forward for your inputs.
>
> --Thanks
> --Kishore
>
> -----Original Message-----
> From: Ignasi Barrera [mailto:nacx@apache.org]
> Sent: Monday, June 12, 2017 11:47 PM
> To: dev@jclouds.apache.org
> Subject: Re: Adding async support for Jclouds library
>
> Hi!
>
> I think that given the output from your benchmarks it is OK to proceed with the async approach. The asnyc and OutputStream versions are not exclusive and the benchmarks show no reason for not starting with implementation of the async feature.
>
> For me, let's go with the next step and open a pull request with the PoC code/continue the discussion/whatever plan you had in mind :)
>
>
> I.
>
> On Jun 12, 2017 4:48 AM, "Battula Kishore" <ba...@adobe.com.invalid>
> wrote:
>
> Hi Andrew, Ignasi
>
> Checking once again any update on this?
>
> -- Thanks
> -- Kishore
>
>
>
>
>
>
>
>
>
> On 06/06/17, 12:10 PM, "Ignasi Barrera" <na...@apache.org> wrote:
>
>>Hi,
>>
>>I can pick it up, but I think we should have some more feedback from
>>Andrew first, at least to see if he still has some concerns about the
>>implementation.
>>I do like the current async approach but it wouldn't be right to move
>>forward without the OK of the team member that had concerns and has
>>been more involved in the design discussion.
>>
>>
>>I,
>>
>>On 6 June 2017 at 08:28, Felix Meschberger <fm...@adobe.com.invalid>
> wrote:
>>> Hi Jclouds dev
>>>
>>> Anyone else who could pick up instead of AndrewG ?
>>>
>>> It would be great to be able to have this support added to JClouds.
>>>
>>> Thanks
>>> Felix
>>>
>>>> Am 06.06.2017 um 05:49 schrieb Andrew Phillips <an...@apache.org>:
>>>>
>>>> Hi Kishore
>>>>
>>>> Andrew G. is travelling for most of June, as far as I understand, so
> will likely be a bit slower to respond. Thanks for your patience!
>>>>
>>>> ap
>>>>
>>>> On 2017-06-05 23:42, Battula Kishore wrote:
>>>>> Hi Andrew,
>>>>> Any update on this?
>>>>> -- Thanks
>>>>> -- Kishore
>>>>> On 29/05/17, 1:06 PM, "Battula Kishore" <ba...@adobe.com.INVALID>
> wrote:
>>>>>> Hi Andrew,
>>>>>> Thanks andrew for the quick response. Here is the GitHub repo and
> instructions on how to run the tests https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fna01.safelinks&data=02%7C01%7C%7C6ad1a11f02da4249a9df08d4b1bf4312%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636328882406496498&sdata=VxmspqySHr7onsxPr75PVxGzQgfEVVJoITrOfy5BdHU%3D&reserved=0.
> protection.outlook.com/?url=https%3A%2F%2Fgithub.com%
> 2Fkishore25kumar%2Fs3proxy-async-test-setup&data=02%7C01%7C%
> 7C8969ed65c6a749aebfa608d4a66583cc%7Cfa7b1b5a7b34438794aed2c178de
> cee1%7C0%7C0%7C636316402308126519&sdata=0YESeCSlmChUOicpe2m9SYC1WaB7m%
> 2B7hIB6o%2BTOp1Jo%3D&reserved=0. The READE.md also has the repo details of s3proxy as well as jclouds implementation. Hope this helps? Let me know if you need anything else?
>>>>>> In the mean time you review these results if you can let me know
>>>>>> the
> design review process I can be prepared for that.
>>>>>> -- Thanks
>>>>>> -- Kishore
>>>>>> On 26/05/17, 12:40 PM, "Andrew Gaul" <ga...@apache.org> wrote:
>>>>>>> Kishore, these are promising results! I reformatted the most
> important
>>>>>>> rows which show a 2x improvement in throughput and latency:
>>>>>>> 10 10,000 Async Http Lib 209 282 48
>>>>>>> 10 10,000 OutputStream 392 542 25
>>>>>>> Can you share the implementation and include instructions on how
>>>>>>> to replicate these tests?
>>>>>>> On Wed, May 24, 2017 at 05:50:52AM +0000, Battula Kishore wrote:
>>>>>>>> Hi,
>>>>>>>> This is Kishore who is working on async poc using mail id(
> kishore25kumar@gmail.com<ma...@gmail.com>). I work at adobe and we wanted to implement async support for jclouds library and contribute it back.
>>>>>>>> From the last discussion I was asked to get the performance
>>>>>>>> numbers
> for the two approaches.
>>>>>>>> Approach 1: Using Http Async Library Approach 2: Using
>>>>>>>> Outputstream Test setup:
>>>>>>>> 1. Both the s3 proxy server and test runner are running in
> same Docker container in azure west-us region.
>>>>>>>> 2. Azure storage account is also residing in same west-us
> region.
>>>>>>>> 3. A bucket is prepopulated with 100,000 files, each file of
> 1 MB size before test start.
>>>>>>>> 4. The test runner sends unique requests to s3proxy to
> download files.
>>>>>>>> Virtual Machine spec: CPU - 8 cores, Memory - 28 GB (Standard_D4
> Azure machine)
>>>>>>>> S3proxy is running with 1 jetty worker thread in all the scenarios.
> The payload size used is 1 MB file. Here are the performance numbers.
>>>>>>>> Test Runner Threads
>>>>>>>> Iteration Per thread
>>>>>>>> Approach
>>>>>>>> Avg response time (ms)
>>>>>>>> 99%tile time (ms)
>>>>>>>> Throughput
>>>>>>>> (Requests / sec)
>>>>>>>> 1
>>>>>>>> 10,000
>>>>>>>> Async Http Lib
>>>>>>>> 45
>>>>>>>> 87
>>>>>>>> 22
>>>>>>>> 5
>>>>>>>> 10,000
>>>>>>>> Async Http Lib
>>>>>>>> 107
>>>>>>>> 159
>>>>>>>> 47
>>>>>>>> 10
>>>>>>>> 10,000
>>>>>>>> Async Http Lib
>>>>>>>> 209
>>>>>>>> 282
>>>>>>>> 48
>>>>>>>> 1
>>>>>>>> 10,000
>>>>>>>> OutputStream
>>>>>>>> 41
>>>>>>>> 85
>>>>>>>> 24
>>>>>>>> 5
>>>>>>>> 10,000
>>>>>>>> OutputStream
>>>>>>>> 190
>>>>>>>> 283
>>>>>>>> 26
>>>>>>>> 10
>>>>>>>> 10,000
>>>>>>>> OutputStream
>>>>>>>> 392
>>>>>>>> 542
>>>>>>>> 25
>>>>>>>> Summary: Under load Http Async Library approach is providing
>>>>>>>> more
> throughput compared to Output stream approach.
>>>>>>>> Both the approaches improve performance. The output stream
>>>>>>>> approach
> can be used along with Http Async library approach which is giving around
> (3-5 ms) improvement in latency.
>>>>>>>> Each approach is independent development. At this point I am
>>>>>>>> keen
> to take up Http Async Library development.
>>>>>>>> -- Thanks
>>>>>>>> -- Kishore
>>>>>>> --
>>>>>>> Andrew Gaul
>>>>>>> https://na01.safelinks.protection.outlook.com/?url=
> http%3A%2F%2Fgaul.org%2F&data=02%7C01%7C%7C8690ee3e00bc4e3da0e608d4a406565e%
> 7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636313794511467457&sdata=XQ%
> 2FshVjdqC3KiVEuyH6%2FJvmDN5DHBmS0kIBx98V89KY%3D&reserved=0
>>>
RE: Adding async support for Jclouds library
Posted by Battula Kishore <ba...@adobe.com.INVALID>.
Hi Ignasi,
Just to do a quick walkthrough what I have already did towards the async approach.
Here is the initial pull request for AsyncPOC
https://github.com/SpandanThakur/jclouds/pull/2
This pull request has gone through couple of reviews from the community members and I have made necessary changes. One last thing I was asked is to implement AsyncBlobInterface instead of adding it to existing BlobInterface. I have done it in this pull request.
https://github.com/SpandanThakur/jclouds/pull/4
I am looking for some inputs on how to proceed from here to get it into a state where it can be merged. Then after the I can follow the same steps for other API's as well once I get familiar with the process for one API. Looking forward for your inputs.
--Thanks
--Kishore
-----Original Message-----
From: Ignasi Barrera [mailto:nacx@apache.org]
Sent: Monday, June 12, 2017 11:47 PM
To: dev@jclouds.apache.org
Subject: Re: Adding async support for Jclouds library
Hi!
I think that given the output from your benchmarks it is OK to proceed with the async approach. The asnyc and OutputStream versions are not exclusive and the benchmarks show no reason for not starting with implementation of the async feature.
For me, let's go with the next step and open a pull request with the PoC code/continue the discussion/whatever plan you had in mind :)
I.
On Jun 12, 2017 4:48 AM, "Battula Kishore" <ba...@adobe.com.invalid>
wrote:
Hi Andrew, Ignasi
Checking once again any update on this?
-- Thanks
-- Kishore
On 06/06/17, 12:10 PM, "Ignasi Barrera" <na...@apache.org> wrote:
>Hi,
>
>I can pick it up, but I think we should have some more feedback from
>Andrew first, at least to see if he still has some concerns about the
>implementation.
>I do like the current async approach but it wouldn't be right to move
>forward without the OK of the team member that had concerns and has
>been more involved in the design discussion.
>
>
>I,
>
>On 6 June 2017 at 08:28, Felix Meschberger <fm...@adobe.com.invalid>
wrote:
>> Hi Jclouds dev
>>
>> Anyone else who could pick up instead of AndrewG ?
>>
>> It would be great to be able to have this support added to JClouds.
>>
>> Thanks
>> Felix
>>
>>> Am 06.06.2017 um 05:49 schrieb Andrew Phillips <an...@apache.org>:
>>>
>>> Hi Kishore
>>>
>>> Andrew G. is travelling for most of June, as far as I understand, so
will likely be a bit slower to respond. Thanks for your patience!
>>>
>>> ap
>>>
>>> On 2017-06-05 23:42, Battula Kishore wrote:
>>>> Hi Andrew,
>>>> Any update on this?
>>>> -- Thanks
>>>> -- Kishore
>>>> On 29/05/17, 1:06 PM, "Battula Kishore" <ba...@adobe.com.INVALID>
wrote:
>>>>> Hi Andrew,
>>>>> Thanks andrew for the quick response. Here is the GitHub repo and
instructions on how to run the tests https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fna01.safelinks&data=02%7C01%7C%7C6ad1a11f02da4249a9df08d4b1bf4312%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636328882406496498&sdata=VxmspqySHr7onsxPr75PVxGzQgfEVVJoITrOfy5BdHU%3D&reserved=0.
protection.outlook.com/?url=https%3A%2F%2Fgithub.com%
2Fkishore25kumar%2Fs3proxy-async-test-setup&data=02%7C01%7C%
7C8969ed65c6a749aebfa608d4a66583cc%7Cfa7b1b5a7b34438794aed2c178de
cee1%7C0%7C0%7C636316402308126519&sdata=0YESeCSlmChUOicpe2m9SYC1WaB7m%
2B7hIB6o%2BTOp1Jo%3D&reserved=0. The READE.md also has the repo details of s3proxy as well as jclouds implementation. Hope this helps? Let me know if you need anything else?
>>>>> In the mean time you review these results if you can let me know
>>>>> the
design review process I can be prepared for that.
>>>>> -- Thanks
>>>>> -- Kishore
>>>>> On 26/05/17, 12:40 PM, "Andrew Gaul" <ga...@apache.org> wrote:
>>>>>> Kishore, these are promising results! I reformatted the most
important
>>>>>> rows which show a 2x improvement in throughput and latency:
>>>>>> 10 10,000 Async Http Lib 209 282 48
>>>>>> 10 10,000 OutputStream 392 542 25
>>>>>> Can you share the implementation and include instructions on how
>>>>>> to replicate these tests?
>>>>>> On Wed, May 24, 2017 at 05:50:52AM +0000, Battula Kishore wrote:
>>>>>>> Hi,
>>>>>>> This is Kishore who is working on async poc using mail id(
kishore25kumar@gmail.com<ma...@gmail.com>). I work at adobe and we wanted to implement async support for jclouds library and contribute it back.
>>>>>>> From the last discussion I was asked to get the performance
>>>>>>> numbers
for the two approaches.
>>>>>>> Approach 1: Using Http Async Library Approach 2: Using
>>>>>>> Outputstream Test setup:
>>>>>>> 1. Both the s3 proxy server and test runner are running in
same Docker container in azure west-us region.
>>>>>>> 2. Azure storage account is also residing in same west-us
region.
>>>>>>> 3. A bucket is prepopulated with 100,000 files, each file of
1 MB size before test start.
>>>>>>> 4. The test runner sends unique requests to s3proxy to
download files.
>>>>>>> Virtual Machine spec: CPU - 8 cores, Memory - 28 GB (Standard_D4
Azure machine)
>>>>>>> S3proxy is running with 1 jetty worker thread in all the scenarios.
The payload size used is 1 MB file. Here are the performance numbers.
>>>>>>> Test Runner Threads
>>>>>>> Iteration Per thread
>>>>>>> Approach
>>>>>>> Avg response time (ms)
>>>>>>> 99%tile time (ms)
>>>>>>> Throughput
>>>>>>> (Requests / sec)
>>>>>>> 1
>>>>>>> 10,000
>>>>>>> Async Http Lib
>>>>>>> 45
>>>>>>> 87
>>>>>>> 22
>>>>>>> 5
>>>>>>> 10,000
>>>>>>> Async Http Lib
>>>>>>> 107
>>>>>>> 159
>>>>>>> 47
>>>>>>> 10
>>>>>>> 10,000
>>>>>>> Async Http Lib
>>>>>>> 209
>>>>>>> 282
>>>>>>> 48
>>>>>>> 1
>>>>>>> 10,000
>>>>>>> OutputStream
>>>>>>> 41
>>>>>>> 85
>>>>>>> 24
>>>>>>> 5
>>>>>>> 10,000
>>>>>>> OutputStream
>>>>>>> 190
>>>>>>> 283
>>>>>>> 26
>>>>>>> 10
>>>>>>> 10,000
>>>>>>> OutputStream
>>>>>>> 392
>>>>>>> 542
>>>>>>> 25
>>>>>>> Summary: Under load Http Async Library approach is providing
>>>>>>> more
throughput compared to Output stream approach.
>>>>>>> Both the approaches improve performance. The output stream
>>>>>>> approach
can be used along with Http Async library approach which is giving around
(3-5 ms) improvement in latency.
>>>>>>> Each approach is independent development. At this point I am
>>>>>>> keen
to take up Http Async Library development.
>>>>>>> -- Thanks
>>>>>>> -- Kishore
>>>>>> --
>>>>>> Andrew Gaul
>>>>>> https://na01.safelinks.protection.outlook.com/?url=
http%3A%2F%2Fgaul.org%2F&data=02%7C01%7C%7C8690ee3e00bc4e3da0e608d4a406565e%
7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636313794511467457&sdata=XQ%
2FshVjdqC3KiVEuyH6%2FJvmDN5DHBmS0kIBx98V89KY%3D&reserved=0
>>
Re: Adding async support for Jclouds library
Posted by Ignasi Barrera <na...@apache.org>.
Hi!
I think that given the output from your benchmarks it is OK to proceed with
the async approach. The asnyc and OutputStream versions are not exclusive
and the benchmarks show no reason for not starting with implementation of
the async feature.
For me, let's go with the next step and open a pull request with the PoC
code/continue the discussion/whatever plan you had in mind :)
I.
On Jun 12, 2017 4:48 AM, "Battula Kishore" <ba...@adobe.com.invalid>
wrote:
Hi Andrew, Ignasi
Checking once again any update on this?
-- Thanks
-- Kishore
On 06/06/17, 12:10 PM, "Ignasi Barrera" <na...@apache.org> wrote:
>Hi,
>
>I can pick it up, but I think we should have some more feedback from
>Andrew first, at least to see if he still has some concerns about the
>implementation.
>I do like the current async approach but it wouldn't be right to move
>forward without the OK of the team member that had concerns and has
>been more involved in the design discussion.
>
>
>I,
>
>On 6 June 2017 at 08:28, Felix Meschberger <fm...@adobe.com.invalid>
wrote:
>> Hi Jclouds dev
>>
>> Anyone else who could pick up instead of AndrewG ?
>>
>> It would be great to be able to have this support added to JClouds.
>>
>> Thanks
>> Felix
>>
>>> Am 06.06.2017 um 05:49 schrieb Andrew Phillips <an...@apache.org>:
>>>
>>> Hi Kishore
>>>
>>> Andrew G. is travelling for most of June, as far as I understand, so
will likely be a bit slower to respond. Thanks for your patience!
>>>
>>> ap
>>>
>>> On 2017-06-05 23:42, Battula Kishore wrote:
>>>> Hi Andrew,
>>>> Any update on this?
>>>> -- Thanks
>>>> -- Kishore
>>>> On 29/05/17, 1:06 PM, "Battula Kishore" <ba...@adobe.com.INVALID>
wrote:
>>>>> Hi Andrew,
>>>>> Thanks andrew for the quick response. Here is the GitHub repo and
instructions on how to run the tests https://na01.safelinks.
protection.outlook.com/?url=https%3A%2F%2Fgithub.com%
2Fkishore25kumar%2Fs3proxy-async-test-setup&data=02%7C01%7C%
7C8969ed65c6a749aebfa608d4a66583cc%7Cfa7b1b5a7b34438794aed2c178de
cee1%7C0%7C0%7C636316402308126519&sdata=0YESeCSlmChUOicpe2m9SYC1WaB7m%
2B7hIB6o%2BTOp1Jo%3D&reserved=0. The READE.md also has the repo details of
s3proxy as well as jclouds implementation. Hope this helps? Let me know if
you need anything else?
>>>>> In the mean time you review these results if you can let me know the
design review process I can be prepared for that.
>>>>> -- Thanks
>>>>> -- Kishore
>>>>> On 26/05/17, 12:40 PM, "Andrew Gaul" <ga...@apache.org> wrote:
>>>>>> Kishore, these are promising results! I reformatted the most
important
>>>>>> rows which show a 2x improvement in throughput and latency:
>>>>>> 10 10,000 Async Http Lib 209 282 48
>>>>>> 10 10,000 OutputStream 392 542 25
>>>>>> Can you share the implementation and include instructions on how to
>>>>>> replicate these tests?
>>>>>> On Wed, May 24, 2017 at 05:50:52AM +0000, Battula Kishore wrote:
>>>>>>> Hi,
>>>>>>> This is Kishore who is working on async poc using mail id(
kishore25kumar@gmail.com<ma...@gmail.com>). I work at adobe
and we wanted to implement async support for jclouds library and contribute
it back.
>>>>>>> From the last discussion I was asked to get the performance numbers
for the two approaches.
>>>>>>> Approach 1: Using Http Async Library
>>>>>>> Approach 2: Using Outputstream
>>>>>>> Test setup:
>>>>>>> 1. Both the s3 proxy server and test runner are running in
same Docker container in azure west-us region.
>>>>>>> 2. Azure storage account is also residing in same west-us
region.
>>>>>>> 3. A bucket is prepopulated with 100,000 files, each file of
1 MB size before test start.
>>>>>>> 4. The test runner sends unique requests to s3proxy to
download files.
>>>>>>> Virtual Machine spec: CPU - 8 cores, Memory - 28 GB (Standard_D4
Azure machine)
>>>>>>> S3proxy is running with 1 jetty worker thread in all the scenarios.
The payload size used is 1 MB file. Here are the performance numbers.
>>>>>>> Test Runner Threads
>>>>>>> Iteration Per thread
>>>>>>> Approach
>>>>>>> Avg response time (ms)
>>>>>>> 99%tile time (ms)
>>>>>>> Throughput
>>>>>>> (Requests / sec)
>>>>>>> 1
>>>>>>> 10,000
>>>>>>> Async Http Lib
>>>>>>> 45
>>>>>>> 87
>>>>>>> 22
>>>>>>> 5
>>>>>>> 10,000
>>>>>>> Async Http Lib
>>>>>>> 107
>>>>>>> 159
>>>>>>> 47
>>>>>>> 10
>>>>>>> 10,000
>>>>>>> Async Http Lib
>>>>>>> 209
>>>>>>> 282
>>>>>>> 48
>>>>>>> 1
>>>>>>> 10,000
>>>>>>> OutputStream
>>>>>>> 41
>>>>>>> 85
>>>>>>> 24
>>>>>>> 5
>>>>>>> 10,000
>>>>>>> OutputStream
>>>>>>> 190
>>>>>>> 283
>>>>>>> 26
>>>>>>> 10
>>>>>>> 10,000
>>>>>>> OutputStream
>>>>>>> 392
>>>>>>> 542
>>>>>>> 25
>>>>>>> Summary: Under load Http Async Library approach is providing more
throughput compared to Output stream approach.
>>>>>>> Both the approaches improve performance. The output stream approach
can be used along with Http Async library approach which is giving around
(3-5 ms) improvement in latency.
>>>>>>> Each approach is independent development. At this point I am keen
to take up Http Async Library development.
>>>>>>> -- Thanks
>>>>>>> -- Kishore
>>>>>> --
>>>>>> Andrew Gaul
>>>>>> https://na01.safelinks.protection.outlook.com/?url=
http%3A%2F%2Fgaul.org%2F&data=02%7C01%7C%7C8690ee3e00bc4e3da0e608d4a406565e%
7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636313794511467457&sdata=XQ%
2FshVjdqC3KiVEuyH6%2FJvmDN5DHBmS0kIBx98V89KY%3D&reserved=0
>>
Re: Adding async support for Jclouds library
Posted by Battula Kishore <ba...@adobe.com.INVALID>.
Hi Andrew, Ignasi
Checking once again any update on this?
-- Thanks
-- Kishore
On 06/06/17, 12:10 PM, "Ignasi Barrera" <na...@apache.org> wrote:
>Hi,
>
>I can pick it up, but I think we should have some more feedback from
>Andrew first, at least to see if he still has some concerns about the
>implementation.
>I do like the current async approach but it wouldn't be right to move
>forward without the OK of the team member that had concerns and has
>been more involved in the design discussion.
>
>
>I,
>
>On 6 June 2017 at 08:28, Felix Meschberger <fm...@adobe.com.invalid> wrote:
>> Hi Jclouds dev
>>
>> Anyone else who could pick up instead of AndrewG ?
>>
>> It would be great to be able to have this support added to JClouds.
>>
>> Thanks
>> Felix
>>
>>> Am 06.06.2017 um 05:49 schrieb Andrew Phillips <an...@apache.org>:
>>>
>>> Hi Kishore
>>>
>>> Andrew G. is travelling for most of June, as far as I understand, so will likely be a bit slower to respond. Thanks for your patience!
>>>
>>> ap
>>>
>>> On 2017-06-05 23:42, Battula Kishore wrote:
>>>> Hi Andrew,
>>>> Any update on this?
>>>> -- Thanks
>>>> -- Kishore
>>>> On 29/05/17, 1:06 PM, "Battula Kishore" <ba...@adobe.com.INVALID> wrote:
>>>>> Hi Andrew,
>>>>> Thanks andrew for the quick response. Here is the GitHub repo and instructions on how to run the tests https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fkishore25kumar%2Fs3proxy-async-test-setup&data=02%7C01%7C%7C8969ed65c6a749aebfa608d4a66583cc%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636316402308126519&sdata=0YESeCSlmChUOicpe2m9SYC1WaB7m%2B7hIB6o%2BTOp1Jo%3D&reserved=0. The READE.md also has the repo details of s3proxy as well as jclouds implementation. Hope this helps? Let me know if you need anything else?
>>>>> In the mean time you review these results if you can let me know the design review process I can be prepared for that.
>>>>> -- Thanks
>>>>> -- Kishore
>>>>> On 26/05/17, 12:40 PM, "Andrew Gaul" <ga...@apache.org> wrote:
>>>>>> Kishore, these are promising results! I reformatted the most important
>>>>>> rows which show a 2x improvement in throughput and latency:
>>>>>> 10 10,000 Async Http Lib 209 282 48
>>>>>> 10 10,000 OutputStream 392 542 25
>>>>>> Can you share the implementation and include instructions on how to
>>>>>> replicate these tests?
>>>>>> On Wed, May 24, 2017 at 05:50:52AM +0000, Battula Kishore wrote:
>>>>>>> Hi,
>>>>>>> This is Kishore who is working on async poc using mail id(kishore25kumar@gmail.com<ma...@gmail.com>). I work at adobe and we wanted to implement async support for jclouds library and contribute it back.
>>>>>>> From the last discussion I was asked to get the performance numbers for the two approaches.
>>>>>>> Approach 1: Using Http Async Library
>>>>>>> Approach 2: Using Outputstream
>>>>>>> Test setup:
>>>>>>> 1. Both the s3 proxy server and test runner are running in same Docker container in azure west-us region.
>>>>>>> 2. Azure storage account is also residing in same west-us region.
>>>>>>> 3. A bucket is prepopulated with 100,000 files, each file of 1 MB size before test start.
>>>>>>> 4. The test runner sends unique requests to s3proxy to download files.
>>>>>>> Virtual Machine spec: CPU - 8 cores, Memory - 28 GB (Standard_D4 Azure machine)
>>>>>>> S3proxy is running with 1 jetty worker thread in all the scenarios. The payload size used is 1 MB file. Here are the performance numbers.
>>>>>>> Test Runner Threads
>>>>>>> Iteration Per thread
>>>>>>> Approach
>>>>>>> Avg response time (ms)
>>>>>>> 99%tile time (ms)
>>>>>>> Throughput
>>>>>>> (Requests / sec)
>>>>>>> 1
>>>>>>> 10,000
>>>>>>> Async Http Lib
>>>>>>> 45
>>>>>>> 87
>>>>>>> 22
>>>>>>> 5
>>>>>>> 10,000
>>>>>>> Async Http Lib
>>>>>>> 107
>>>>>>> 159
>>>>>>> 47
>>>>>>> 10
>>>>>>> 10,000
>>>>>>> Async Http Lib
>>>>>>> 209
>>>>>>> 282
>>>>>>> 48
>>>>>>> 1
>>>>>>> 10,000
>>>>>>> OutputStream
>>>>>>> 41
>>>>>>> 85
>>>>>>> 24
>>>>>>> 5
>>>>>>> 10,000
>>>>>>> OutputStream
>>>>>>> 190
>>>>>>> 283
>>>>>>> 26
>>>>>>> 10
>>>>>>> 10,000
>>>>>>> OutputStream
>>>>>>> 392
>>>>>>> 542
>>>>>>> 25
>>>>>>> Summary: Under load Http Async Library approach is providing more throughput compared to Output stream approach.
>>>>>>> Both the approaches improve performance. The output stream approach can be used along with Http Async library approach which is giving around (3-5 ms) improvement in latency.
>>>>>>> Each approach is independent development. At this point I am keen to take up Http Async Library development.
>>>>>>> -- Thanks
>>>>>>> -- Kishore
>>>>>> --
>>>>>> Andrew Gaul
>>>>>> https://na01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fgaul.org%2F&data=02%7C01%7C%7C8690ee3e00bc4e3da0e608d4a406565e%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636313794511467457&sdata=XQ%2FshVjdqC3KiVEuyH6%2FJvmDN5DHBmS0kIBx98V89KY%3D&reserved=0
>>
Re: Adding async support for Jclouds library
Posted by Ignasi Barrera <na...@apache.org>.
Hi,
I can pick it up, but I think we should have some more feedback from
Andrew first, at least to see if he still has some concerns about the
implementation.
I do like the current async approach but it wouldn't be right to move
forward without the OK of the team member that had concerns and has
been more involved in the design discussion.
I,
On 6 June 2017 at 08:28, Felix Meschberger <fm...@adobe.com.invalid> wrote:
> Hi Jclouds dev
>
> Anyone else who could pick up instead of AndrewG ?
>
> It would be great to be able to have this support added to JClouds.
>
> Thanks
> Felix
>
>> Am 06.06.2017 um 05:49 schrieb Andrew Phillips <an...@apache.org>:
>>
>> Hi Kishore
>>
>> Andrew G. is travelling for most of June, as far as I understand, so will likely be a bit slower to respond. Thanks for your patience!
>>
>> ap
>>
>> On 2017-06-05 23:42, Battula Kishore wrote:
>>> Hi Andrew,
>>> Any update on this?
>>> -- Thanks
>>> -- Kishore
>>> On 29/05/17, 1:06 PM, "Battula Kishore" <ba...@adobe.com.INVALID> wrote:
>>>> Hi Andrew,
>>>> Thanks andrew for the quick response. Here is the GitHub repo and instructions on how to run the tests https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fkishore25kumar%2Fs3proxy-async-test-setup&data=02%7C01%7C%7C8969ed65c6a749aebfa608d4a66583cc%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636316402308126519&sdata=0YESeCSlmChUOicpe2m9SYC1WaB7m%2B7hIB6o%2BTOp1Jo%3D&reserved=0. The READE.md also has the repo details of s3proxy as well as jclouds implementation. Hope this helps? Let me know if you need anything else?
>>>> In the mean time you review these results if you can let me know the design review process I can be prepared for that.
>>>> -- Thanks
>>>> -- Kishore
>>>> On 26/05/17, 12:40 PM, "Andrew Gaul" <ga...@apache.org> wrote:
>>>>> Kishore, these are promising results! I reformatted the most important
>>>>> rows which show a 2x improvement in throughput and latency:
>>>>> 10 10,000 Async Http Lib 209 282 48
>>>>> 10 10,000 OutputStream 392 542 25
>>>>> Can you share the implementation and include instructions on how to
>>>>> replicate these tests?
>>>>> On Wed, May 24, 2017 at 05:50:52AM +0000, Battula Kishore wrote:
>>>>>> Hi,
>>>>>> This is Kishore who is working on async poc using mail id(kishore25kumar@gmail.com<ma...@gmail.com>). I work at adobe and we wanted to implement async support for jclouds library and contribute it back.
>>>>>> From the last discussion I was asked to get the performance numbers for the two approaches.
>>>>>> Approach 1: Using Http Async Library
>>>>>> Approach 2: Using Outputstream
>>>>>> Test setup:
>>>>>> 1. Both the s3 proxy server and test runner are running in same Docker container in azure west-us region.
>>>>>> 2. Azure storage account is also residing in same west-us region.
>>>>>> 3. A bucket is prepopulated with 100,000 files, each file of 1 MB size before test start.
>>>>>> 4. The test runner sends unique requests to s3proxy to download files.
>>>>>> Virtual Machine spec: CPU - 8 cores, Memory - 28 GB (Standard_D4 Azure machine)
>>>>>> S3proxy is running with 1 jetty worker thread in all the scenarios. The payload size used is 1 MB file. Here are the performance numbers.
>>>>>> Test Runner Threads
>>>>>> Iteration Per thread
>>>>>> Approach
>>>>>> Avg response time (ms)
>>>>>> 99%tile time (ms)
>>>>>> Throughput
>>>>>> (Requests / sec)
>>>>>> 1
>>>>>> 10,000
>>>>>> Async Http Lib
>>>>>> 45
>>>>>> 87
>>>>>> 22
>>>>>> 5
>>>>>> 10,000
>>>>>> Async Http Lib
>>>>>> 107
>>>>>> 159
>>>>>> 47
>>>>>> 10
>>>>>> 10,000
>>>>>> Async Http Lib
>>>>>> 209
>>>>>> 282
>>>>>> 48
>>>>>> 1
>>>>>> 10,000
>>>>>> OutputStream
>>>>>> 41
>>>>>> 85
>>>>>> 24
>>>>>> 5
>>>>>> 10,000
>>>>>> OutputStream
>>>>>> 190
>>>>>> 283
>>>>>> 26
>>>>>> 10
>>>>>> 10,000
>>>>>> OutputStream
>>>>>> 392
>>>>>> 542
>>>>>> 25
>>>>>> Summary: Under load Http Async Library approach is providing more throughput compared to Output stream approach.
>>>>>> Both the approaches improve performance. The output stream approach can be used along with Http Async library approach which is giving around (3-5 ms) improvement in latency.
>>>>>> Each approach is independent development. At this point I am keen to take up Http Async Library development.
>>>>>> -- Thanks
>>>>>> -- Kishore
>>>>> --
>>>>> Andrew Gaul
>>>>> https://na01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fgaul.org%2F&data=02%7C01%7C%7C8690ee3e00bc4e3da0e608d4a406565e%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636313794511467457&sdata=XQ%2FshVjdqC3KiVEuyH6%2FJvmDN5DHBmS0kIBx98V89KY%3D&reserved=0
>
RE: Adding async support for Jclouds library
Posted by Battula Kishore <ba...@adobe.com.INVALID>.
Hi Andrew,
It's been very long time since I published the performance number. Is there any update on this?
--Thanks
--Kishore
-----Original Message-----
From: Battula Kishore [mailto:battula@adobe.com.INVALID]
Sent: Wednesday, July 26, 2017 10:59 AM
To: dev@jclouds.apache.org
Subject: RE: Adding async support for Jclouds library
Hi Andrew,
Did you get a chance to look into the performance numbers I have published? We are blocked on this.
--Thanks
--Kishore
-----Original Message-----
From: Battula Kishore [mailto:battula@adobe.com.INVALID]
Sent: Sunday, July 23, 2017 10:13 AM
To: dev@jclouds.apache.org
Subject: Re: Adding async support for Jclouds library
Hi Andrew,
Did you get a chance to look into the performance numbers I have published?
-- Thanks
-- Kishore
On 17/07/17, 11:08 AM, "Battula Kishore" <ba...@adobe.com.INVALID> wrote:
Hi Andrew,
Did you get a chance to look into the async implementation and numbers I have published?
-- Thanks
-- Kishore
On 24/06/17, 7:19 AM, "Battula Kishore" <ba...@adobe.com.INVALID> wrote:
Hi Andrew,
I can totally understand the concerns you have. Please let me know when you can start work on this? To reiterate this is one of the important feature for us and want to use it in our production systems. We wanted to contribute back the async changes to open source.
-- Thanks
-- Kishore
On 22/06/17, 12:00 AM, "Dancy, Chris" <Ch...@pega.com> wrote:
>Hey Everyone,
>
>I'm generally just a lurker on this list but we use jclouds internally here for various things. I've been following the async work and it sounds promising. If you need additional testing done please send around the details on how to do so and we'd be more than happy to help out and provide more benchmarks.
>
>Thanks,
>Chris
>________________________________________
>From: Andrew Gaul <ga...@apache.org>
>Sent: Wednesday, June 21, 2017 2:12 PM
>To: dev@jclouds.apache.org
>Subject: Re: Adding async support for Jclouds library
>
>As Andrew Philips said, I have little availability until early July.
>While others have capability of reviewing these changes, I prefer to
>look at them as well since we have screwed up a previous async API. I
>continue to doubt the async benchmark results and I want to further
>understand what you have measured. Given that we have been talking
>about these changes since March and earlier responses had no particular
>urgency, including ignoring my mails, we should not have any such
>urgency now. Assuming this is still open, I look forward to working on
>it in a few weeks.
>
>On Tue, Jun 06, 2017 at 06:28:06AM +0000, Felix Meschberger wrote:
>> Hi Jclouds dev
>>
>> Anyone else who could pick up instead of AndrewG ?
>>
>> It would be great to be able to have this support added to JClouds.
>>
>> Thanks
>> Felix
>>
>> > Am 06.06.2017 um 05:49 schrieb Andrew Phillips <an...@apache.org>:
>> >
>> > Hi Kishore
>> >
>> > Andrew G. is travelling for most of June, as far as I understand, so will likely be a bit slower to respond. Thanks for your patience!
>> >
>> > ap
>> >
>> > On 2017-06-05 23:42, Battula Kishore wrote:
>> >> Hi Andrew,
>> >> Any update on this?
>> >> -- Thanks
>> >> -- Kishore
>> >> On 29/05/17, 1:06 PM, "Battula Kishore" <ba...@adobe.com.INVALID> wrote:
>> >>> Hi Andrew,
>> >>> Thanks andrew for the quick response. Here is the GitHub repo and instructions on how to run the tests https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fkishore25kumar%2Fs3proxy-async-test-setup&data=02%7C01%7C%7C8969ed65c6a749aebfa608d4a66583cc%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636316402308126519&sdata=0YESeCSlmChUOicpe2m9SYC1WaB7m%2B7hIB6o%2BTOp1Jo%3D&reserved=0. The READE.md also has the repo details of s3proxy as well as jclouds implementation. Hope this helps? Let me know if you need anything else?
>> >>> In the mean time you review these results if you can let me know the design review process I can be prepared for that.
>> >>> -- Thanks
>> >>> -- Kishore
>> >>> On 26/05/17, 12:40 PM, "Andrew Gaul" <ga...@apache.org> wrote:
>> >>>> Kishore, these are promising results! I reformatted the most important
>> >>>> rows which show a 2x improvement in throughput and latency:
>> >>>> 10 10,000 Async Http Lib 209 282 48
>> >>>> 10 10,000 OutputStream 392 542 25
>> >>>> Can you share the implementation and include instructions on how to
>> >>>> replicate these tests?
>> >>>> On Wed, May 24, 2017 at 05:50:52AM +0000, Battula Kishore wrote:
>> >>>>> Hi,
>> >>>>> This is Kishore who is working on async poc using mail id(kishore25kumar@gmail.com<ma...@gmail.com>). I work at adobe and we wanted to implement async support for jclouds library and contribute it back.
>> >>>>> From the last discussion I was asked to get the performance numbers for the two approaches.
>> >>>>> Approach 1: Using Http Async Library
>> >>>>> Approach 2: Using Outputstream
>> >>>>> Test setup:
>> >>>>> 1. Both the s3 proxy server and test runner are running in same Docker container in azure west-us region.
>> >>>>> 2. Azure storage account is also residing in same west-us region.
>> >>>>> 3. A bucket is prepopulated with 100,000 files, each file of 1 MB size before test start.
>> >>>>> 4. The test runner sends unique requests to s3proxy to download files.
>> >>>>> Virtual Machine spec: CPU - 8 cores, Memory - 28 GB (Standard_D4 Azure machine)
>> >>>>> S3proxy is running with 1 jetty worker thread in all the scenarios. The payload size used is 1 MB file. Here are the performance numbers.
>> >>>>> Test Runner Threads
>> >>>>> Iteration Per thread
>> >>>>> Approach
>> >>>>> Avg response time (ms)
>> >>>>> 99%tile time (ms)
>> >>>>> Throughput
>> >>>>> (Requests / sec)
>> >>>>> 1
>> >>>>> 10,000
>> >>>>> Async Http Lib
>> >>>>> 45
>> >>>>> 87
>> >>>>> 22
>> >>>>> 5
>> >>>>> 10,000
>> >>>>> Async Http Lib
>> >>>>> 107
>> >>>>> 159
>> >>>>> 47
>> >>>>> 10
>> >>>>> 10,000
>> >>>>> Async Http Lib
>> >>>>> 209
>> >>>>> 282
>> >>>>> 48
>> >>>>> 1
>> >>>>> 10,000
>> >>>>> OutputStream
>> >>>>> 41
>> >>>>> 85
>> >>>>> 24
>> >>>>> 5
>> >>>>> 10,000
>> >>>>> OutputStream
>> >>>>> 190
>> >>>>> 283
>> >>>>> 26
>> >>>>> 10
>> >>>>> 10,000
>> >>>>> OutputStream
>> >>>>> 392
>> >>>>> 542
>> >>>>> 25
>> >>>>> Summary: Under load Http Async Library approach is providing more throughput compared to Output stream approach.
>> >>>>> Both the approaches improve performance. The output stream approach can be used along with Http Async library approach which is giving around (3-5 ms) improvement in latency.
>> >>>>> Each approach is independent development. At this point I am keen to take up Http Async Library development.
>> >>>>> -- Thanks
>> >>>>> -- Kishore
>> >>>> --
>> >>>> Andrew Gaul
>> >>>> https://na01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fgaul.org%2F&data=02%7C01%7C%7C8690ee3e00bc4e3da0e608d4a406565e%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636313794511467457&sdata=XQ%2FshVjdqC3KiVEuyH6%2FJvmDN5DHBmS0kIBx98V89KY%3D&reserved=0
>>
>
>--
>Andrew Gaul
>https://na01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fgaul.org%2F&data=02%7C01%7C%7Cccc73352a8964b931b6a08d4b8d392a3%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636336666213699146&sdata=58ycimrkelgVaMn69lHY2nWE2qlzK6gpkuH%2FZP06M6E%3D&reserved=0
>
>
>
RE: Adding async support for Jclouds library
Posted by Battula Kishore <ba...@adobe.com.INVALID>.
Hi Andrew,
Did you get a chance to look into the performance numbers I have published? We are blocked on this.
--Thanks
--Kishore
-----Original Message-----
From: Battula Kishore [mailto:battula@adobe.com.INVALID]
Sent: Sunday, July 23, 2017 10:13 AM
To: dev@jclouds.apache.org
Subject: Re: Adding async support for Jclouds library
Hi Andrew,
Did you get a chance to look into the performance numbers I have published?
-- Thanks
-- Kishore
On 17/07/17, 11:08 AM, "Battula Kishore" <ba...@adobe.com.INVALID> wrote:
Hi Andrew,
Did you get a chance to look into the async implementation and numbers I have published?
-- Thanks
-- Kishore
On 24/06/17, 7:19 AM, "Battula Kishore" <ba...@adobe.com.INVALID> wrote:
Hi Andrew,
I can totally understand the concerns you have. Please let me know when you can start work on this? To reiterate this is one of the important feature for us and want to use it in our production systems. We wanted to contribute back the async changes to open source.
-- Thanks
-- Kishore
On 22/06/17, 12:00 AM, "Dancy, Chris" <Ch...@pega.com> wrote:
>Hey Everyone,
>
>I'm generally just a lurker on this list but we use jclouds internally here for various things. I've been following the async work and it sounds promising. If you need additional testing done please send around the details on how to do so and we'd be more than happy to help out and provide more benchmarks.
>
>Thanks,
>Chris
>________________________________________
>From: Andrew Gaul <ga...@apache.org>
>Sent: Wednesday, June 21, 2017 2:12 PM
>To: dev@jclouds.apache.org
>Subject: Re: Adding async support for Jclouds library
>
>As Andrew Philips said, I have little availability until early July.
>While others have capability of reviewing these changes, I prefer to
>look at them as well since we have screwed up a previous async API. I
>continue to doubt the async benchmark results and I want to further
>understand what you have measured. Given that we have been talking
>about these changes since March and earlier responses had no particular
>urgency, including ignoring my mails, we should not have any such
>urgency now. Assuming this is still open, I look forward to working on
>it in a few weeks.
>
>On Tue, Jun 06, 2017 at 06:28:06AM +0000, Felix Meschberger wrote:
>> Hi Jclouds dev
>>
>> Anyone else who could pick up instead of AndrewG ?
>>
>> It would be great to be able to have this support added to JClouds.
>>
>> Thanks
>> Felix
>>
>> > Am 06.06.2017 um 05:49 schrieb Andrew Phillips <an...@apache.org>:
>> >
>> > Hi Kishore
>> >
>> > Andrew G. is travelling for most of June, as far as I understand, so will likely be a bit slower to respond. Thanks for your patience!
>> >
>> > ap
>> >
>> > On 2017-06-05 23:42, Battula Kishore wrote:
>> >> Hi Andrew,
>> >> Any update on this?
>> >> -- Thanks
>> >> -- Kishore
>> >> On 29/05/17, 1:06 PM, "Battula Kishore" <ba...@adobe.com.INVALID> wrote:
>> >>> Hi Andrew,
>> >>> Thanks andrew for the quick response. Here is the GitHub repo and instructions on how to run the tests https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fkishore25kumar%2Fs3proxy-async-test-setup&data=02%7C01%7C%7C8969ed65c6a749aebfa608d4a66583cc%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636316402308126519&sdata=0YESeCSlmChUOicpe2m9SYC1WaB7m%2B7hIB6o%2BTOp1Jo%3D&reserved=0. The READE.md also has the repo details of s3proxy as well as jclouds implementation. Hope this helps? Let me know if you need anything else?
>> >>> In the mean time you review these results if you can let me know the design review process I can be prepared for that.
>> >>> -- Thanks
>> >>> -- Kishore
>> >>> On 26/05/17, 12:40 PM, "Andrew Gaul" <ga...@apache.org> wrote:
>> >>>> Kishore, these are promising results! I reformatted the most important
>> >>>> rows which show a 2x improvement in throughput and latency:
>> >>>> 10 10,000 Async Http Lib 209 282 48
>> >>>> 10 10,000 OutputStream 392 542 25
>> >>>> Can you share the implementation and include instructions on how to
>> >>>> replicate these tests?
>> >>>> On Wed, May 24, 2017 at 05:50:52AM +0000, Battula Kishore wrote:
>> >>>>> Hi,
>> >>>>> This is Kishore who is working on async poc using mail id(kishore25kumar@gmail.com<ma...@gmail.com>). I work at adobe and we wanted to implement async support for jclouds library and contribute it back.
>> >>>>> From the last discussion I was asked to get the performance numbers for the two approaches.
>> >>>>> Approach 1: Using Http Async Library
>> >>>>> Approach 2: Using Outputstream
>> >>>>> Test setup:
>> >>>>> 1. Both the s3 proxy server and test runner are running in same Docker container in azure west-us region.
>> >>>>> 2. Azure storage account is also residing in same west-us region.
>> >>>>> 3. A bucket is prepopulated with 100,000 files, each file of 1 MB size before test start.
>> >>>>> 4. The test runner sends unique requests to s3proxy to download files.
>> >>>>> Virtual Machine spec: CPU - 8 cores, Memory - 28 GB (Standard_D4 Azure machine)
>> >>>>> S3proxy is running with 1 jetty worker thread in all the scenarios. The payload size used is 1 MB file. Here are the performance numbers.
>> >>>>> Test Runner Threads
>> >>>>> Iteration Per thread
>> >>>>> Approach
>> >>>>> Avg response time (ms)
>> >>>>> 99%tile time (ms)
>> >>>>> Throughput
>> >>>>> (Requests / sec)
>> >>>>> 1
>> >>>>> 10,000
>> >>>>> Async Http Lib
>> >>>>> 45
>> >>>>> 87
>> >>>>> 22
>> >>>>> 5
>> >>>>> 10,000
>> >>>>> Async Http Lib
>> >>>>> 107
>> >>>>> 159
>> >>>>> 47
>> >>>>> 10
>> >>>>> 10,000
>> >>>>> Async Http Lib
>> >>>>> 209
>> >>>>> 282
>> >>>>> 48
>> >>>>> 1
>> >>>>> 10,000
>> >>>>> OutputStream
>> >>>>> 41
>> >>>>> 85
>> >>>>> 24
>> >>>>> 5
>> >>>>> 10,000
>> >>>>> OutputStream
>> >>>>> 190
>> >>>>> 283
>> >>>>> 26
>> >>>>> 10
>> >>>>> 10,000
>> >>>>> OutputStream
>> >>>>> 392
>> >>>>> 542
>> >>>>> 25
>> >>>>> Summary: Under load Http Async Library approach is providing more throughput compared to Output stream approach.
>> >>>>> Both the approaches improve performance. The output stream approach can be used along with Http Async library approach which is giving around (3-5 ms) improvement in latency.
>> >>>>> Each approach is independent development. At this point I am keen to take up Http Async Library development.
>> >>>>> -- Thanks
>> >>>>> -- Kishore
>> >>>> --
>> >>>> Andrew Gaul
>> >>>> https://na01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fgaul.org%2F&data=02%7C01%7C%7C8690ee3e00bc4e3da0e608d4a406565e%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636313794511467457&sdata=XQ%2FshVjdqC3KiVEuyH6%2FJvmDN5DHBmS0kIBx98V89KY%3D&reserved=0
>>
>
>--
>Andrew Gaul
>https://na01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fgaul.org%2F&data=02%7C01%7C%7Cccc73352a8964b931b6a08d4b8d392a3%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636336666213699146&sdata=58ycimrkelgVaMn69lHY2nWE2qlzK6gpkuH%2FZP06M6E%3D&reserved=0
>
>
>
Re: Adding async support for Jclouds library
Posted by Battula Kishore <ba...@adobe.com.INVALID>.
Hi Andrew,
Did you get a chance to look into the performance numbers I have published?
-- Thanks
-- Kishore
On 17/07/17, 11:08 AM, "Battula Kishore" <ba...@adobe.com.INVALID> wrote:
Hi Andrew,
Did you get a chance to look into the async implementation and numbers I have published?
-- Thanks
-- Kishore
On 24/06/17, 7:19 AM, "Battula Kishore" <ba...@adobe.com.INVALID> wrote:
Hi Andrew,
I can totally understand the concerns you have. Please let me know when you can start work on this? To reiterate this is one of the important feature for us and want to use it in our production systems. We wanted to contribute back the async changes to open source.
-- Thanks
-- Kishore
On 22/06/17, 12:00 AM, "Dancy, Chris" <Ch...@pega.com> wrote:
>Hey Everyone,
>
>I'm generally just a lurker on this list but we use jclouds internally here for various things. I've been following the async work and it sounds promising. If you need additional testing done please send around the details on how to do so and we'd be more than happy to help out and provide more benchmarks.
>
>Thanks,
>Chris
>________________________________________
>From: Andrew Gaul <ga...@apache.org>
>Sent: Wednesday, June 21, 2017 2:12 PM
>To: dev@jclouds.apache.org
>Subject: Re: Adding async support for Jclouds library
>
>As Andrew Philips said, I have little availability until early July.
>While others have capability of reviewing these changes, I prefer to
>look at them as well since we have screwed up a previous async API. I
>continue to doubt the async benchmark results and I want to further
>understand what you have measured. Given that we have been talking
>about these changes since March and earlier responses had no particular
>urgency, including ignoring my mails, we should not have any such
>urgency now. Assuming this is still open, I look forward to working on
>it in a few weeks.
>
>On Tue, Jun 06, 2017 at 06:28:06AM +0000, Felix Meschberger wrote:
>> Hi Jclouds dev
>>
>> Anyone else who could pick up instead of AndrewG ?
>>
>> It would be great to be able to have this support added to JClouds.
>>
>> Thanks
>> Felix
>>
>> > Am 06.06.2017 um 05:49 schrieb Andrew Phillips <an...@apache.org>:
>> >
>> > Hi Kishore
>> >
>> > Andrew G. is travelling for most of June, as far as I understand, so will likely be a bit slower to respond. Thanks for your patience!
>> >
>> > ap
>> >
>> > On 2017-06-05 23:42, Battula Kishore wrote:
>> >> Hi Andrew,
>> >> Any update on this?
>> >> -- Thanks
>> >> -- Kishore
>> >> On 29/05/17, 1:06 PM, "Battula Kishore" <ba...@adobe.com.INVALID> wrote:
>> >>> Hi Andrew,
>> >>> Thanks andrew for the quick response. Here is the GitHub repo and instructions on how to run the tests https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fkishore25kumar%2Fs3proxy-async-test-setup&data=02%7C01%7C%7C8969ed65c6a749aebfa608d4a66583cc%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636316402308126519&sdata=0YESeCSlmChUOicpe2m9SYC1WaB7m%2B7hIB6o%2BTOp1Jo%3D&reserved=0. The READE.md also has the repo details of s3proxy as well as jclouds implementation. Hope this helps? Let me know if you need anything else?
>> >>> In the mean time you review these results if you can let me know the design review process I can be prepared for that.
>> >>> -- Thanks
>> >>> -- Kishore
>> >>> On 26/05/17, 12:40 PM, "Andrew Gaul" <ga...@apache.org> wrote:
>> >>>> Kishore, these are promising results! I reformatted the most important
>> >>>> rows which show a 2x improvement in throughput and latency:
>> >>>> 10 10,000 Async Http Lib 209 282 48
>> >>>> 10 10,000 OutputStream 392 542 25
>> >>>> Can you share the implementation and include instructions on how to
>> >>>> replicate these tests?
>> >>>> On Wed, May 24, 2017 at 05:50:52AM +0000, Battula Kishore wrote:
>> >>>>> Hi,
>> >>>>> This is Kishore who is working on async poc using mail id(kishore25kumar@gmail.com<ma...@gmail.com>). I work at adobe and we wanted to implement async support for jclouds library and contribute it back.
>> >>>>> From the last discussion I was asked to get the performance numbers for the two approaches.
>> >>>>> Approach 1: Using Http Async Library
>> >>>>> Approach 2: Using Outputstream
>> >>>>> Test setup:
>> >>>>> 1. Both the s3 proxy server and test runner are running in same Docker container in azure west-us region.
>> >>>>> 2. Azure storage account is also residing in same west-us region.
>> >>>>> 3. A bucket is prepopulated with 100,000 files, each file of 1 MB size before test start.
>> >>>>> 4. The test runner sends unique requests to s3proxy to download files.
>> >>>>> Virtual Machine spec: CPU - 8 cores, Memory - 28 GB (Standard_D4 Azure machine)
>> >>>>> S3proxy is running with 1 jetty worker thread in all the scenarios. The payload size used is 1 MB file. Here are the performance numbers.
>> >>>>> Test Runner Threads
>> >>>>> Iteration Per thread
>> >>>>> Approach
>> >>>>> Avg response time (ms)
>> >>>>> 99%tile time (ms)
>> >>>>> Throughput
>> >>>>> (Requests / sec)
>> >>>>> 1
>> >>>>> 10,000
>> >>>>> Async Http Lib
>> >>>>> 45
>> >>>>> 87
>> >>>>> 22
>> >>>>> 5
>> >>>>> 10,000
>> >>>>> Async Http Lib
>> >>>>> 107
>> >>>>> 159
>> >>>>> 47
>> >>>>> 10
>> >>>>> 10,000
>> >>>>> Async Http Lib
>> >>>>> 209
>> >>>>> 282
>> >>>>> 48
>> >>>>> 1
>> >>>>> 10,000
>> >>>>> OutputStream
>> >>>>> 41
>> >>>>> 85
>> >>>>> 24
>> >>>>> 5
>> >>>>> 10,000
>> >>>>> OutputStream
>> >>>>> 190
>> >>>>> 283
>> >>>>> 26
>> >>>>> 10
>> >>>>> 10,000
>> >>>>> OutputStream
>> >>>>> 392
>> >>>>> 542
>> >>>>> 25
>> >>>>> Summary: Under load Http Async Library approach is providing more throughput compared to Output stream approach.
>> >>>>> Both the approaches improve performance. The output stream approach can be used along with Http Async library approach which is giving around (3-5 ms) improvement in latency.
>> >>>>> Each approach is independent development. At this point I am keen to take up Http Async Library development.
>> >>>>> -- Thanks
>> >>>>> -- Kishore
>> >>>> --
>> >>>> Andrew Gaul
>> >>>> https://na01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fgaul.org%2F&data=02%7C01%7C%7C8690ee3e00bc4e3da0e608d4a406565e%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636313794511467457&sdata=XQ%2FshVjdqC3KiVEuyH6%2FJvmDN5DHBmS0kIBx98V89KY%3D&reserved=0
>>
>
>--
>Andrew Gaul
>https://na01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fgaul.org%2F&data=02%7C01%7C%7Cccc73352a8964b931b6a08d4b8d392a3%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636336666213699146&sdata=58ycimrkelgVaMn69lHY2nWE2qlzK6gpkuH%2FZP06M6E%3D&reserved=0
>
>
>
Re: Adding async support for Jclouds library
Posted by Battula Kishore <ba...@adobe.com.INVALID>.
Hi Andrew,
Did you get a chance to look into the async implementation and numbers I have published?
-- Thanks
-- Kishore
On 24/06/17, 7:19 AM, "Battula Kishore" <ba...@adobe.com.INVALID> wrote:
Hi Andrew,
I can totally understand the concerns you have. Please let me know when you can start work on this? To reiterate this is one of the important feature for us and want to use it in our production systems. We wanted to contribute back the async changes to open source.
-- Thanks
-- Kishore
On 22/06/17, 12:00 AM, "Dancy, Chris" <Ch...@pega.com> wrote:
>Hey Everyone,
>
>I'm generally just a lurker on this list but we use jclouds internally here for various things. I've been following the async work and it sounds promising. If you need additional testing done please send around the details on how to do so and we'd be more than happy to help out and provide more benchmarks.
>
>Thanks,
>Chris
>________________________________________
>From: Andrew Gaul <ga...@apache.org>
>Sent: Wednesday, June 21, 2017 2:12 PM
>To: dev@jclouds.apache.org
>Subject: Re: Adding async support for Jclouds library
>
>As Andrew Philips said, I have little availability until early July.
>While others have capability of reviewing these changes, I prefer to
>look at them as well since we have screwed up a previous async API. I
>continue to doubt the async benchmark results and I want to further
>understand what you have measured. Given that we have been talking
>about these changes since March and earlier responses had no particular
>urgency, including ignoring my mails, we should not have any such
>urgency now. Assuming this is still open, I look forward to working on
>it in a few weeks.
>
>On Tue, Jun 06, 2017 at 06:28:06AM +0000, Felix Meschberger wrote:
>> Hi Jclouds dev
>>
>> Anyone else who could pick up instead of AndrewG ?
>>
>> It would be great to be able to have this support added to JClouds.
>>
>> Thanks
>> Felix
>>
>> > Am 06.06.2017 um 05:49 schrieb Andrew Phillips <an...@apache.org>:
>> >
>> > Hi Kishore
>> >
>> > Andrew G. is travelling for most of June, as far as I understand, so will likely be a bit slower to respond. Thanks for your patience!
>> >
>> > ap
>> >
>> > On 2017-06-05 23:42, Battula Kishore wrote:
>> >> Hi Andrew,
>> >> Any update on this?
>> >> -- Thanks
>> >> -- Kishore
>> >> On 29/05/17, 1:06 PM, "Battula Kishore" <ba...@adobe.com.INVALID> wrote:
>> >>> Hi Andrew,
>> >>> Thanks andrew for the quick response. Here is the GitHub repo and instructions on how to run the tests https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fkishore25kumar%2Fs3proxy-async-test-setup&data=02%7C01%7C%7C8969ed65c6a749aebfa608d4a66583cc%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636316402308126519&sdata=0YESeCSlmChUOicpe2m9SYC1WaB7m%2B7hIB6o%2BTOp1Jo%3D&reserved=0. The READE.md also has the repo details of s3proxy as well as jclouds implementation. Hope this helps? Let me know if you need anything else?
>> >>> In the mean time you review these results if you can let me know the design review process I can be prepared for that.
>> >>> -- Thanks
>> >>> -- Kishore
>> >>> On 26/05/17, 12:40 PM, "Andrew Gaul" <ga...@apache.org> wrote:
>> >>>> Kishore, these are promising results! I reformatted the most important
>> >>>> rows which show a 2x improvement in throughput and latency:
>> >>>> 10 10,000 Async Http Lib 209 282 48
>> >>>> 10 10,000 OutputStream 392 542 25
>> >>>> Can you share the implementation and include instructions on how to
>> >>>> replicate these tests?
>> >>>> On Wed, May 24, 2017 at 05:50:52AM +0000, Battula Kishore wrote:
>> >>>>> Hi,
>> >>>>> This is Kishore who is working on async poc using mail id(kishore25kumar@gmail.com<ma...@gmail.com>). I work at adobe and we wanted to implement async support for jclouds library and contribute it back.
>> >>>>> From the last discussion I was asked to get the performance numbers for the two approaches.
>> >>>>> Approach 1: Using Http Async Library
>> >>>>> Approach 2: Using Outputstream
>> >>>>> Test setup:
>> >>>>> 1. Both the s3 proxy server and test runner are running in same Docker container in azure west-us region.
>> >>>>> 2. Azure storage account is also residing in same west-us region.
>> >>>>> 3. A bucket is prepopulated with 100,000 files, each file of 1 MB size before test start.
>> >>>>> 4. The test runner sends unique requests to s3proxy to download files.
>> >>>>> Virtual Machine spec: CPU - 8 cores, Memory - 28 GB (Standard_D4 Azure machine)
>> >>>>> S3proxy is running with 1 jetty worker thread in all the scenarios. The payload size used is 1 MB file. Here are the performance numbers.
>> >>>>> Test Runner Threads
>> >>>>> Iteration Per thread
>> >>>>> Approach
>> >>>>> Avg response time (ms)
>> >>>>> 99%tile time (ms)
>> >>>>> Throughput
>> >>>>> (Requests / sec)
>> >>>>> 1
>> >>>>> 10,000
>> >>>>> Async Http Lib
>> >>>>> 45
>> >>>>> 87
>> >>>>> 22
>> >>>>> 5
>> >>>>> 10,000
>> >>>>> Async Http Lib
>> >>>>> 107
>> >>>>> 159
>> >>>>> 47
>> >>>>> 10
>> >>>>> 10,000
>> >>>>> Async Http Lib
>> >>>>> 209
>> >>>>> 282
>> >>>>> 48
>> >>>>> 1
>> >>>>> 10,000
>> >>>>> OutputStream
>> >>>>> 41
>> >>>>> 85
>> >>>>> 24
>> >>>>> 5
>> >>>>> 10,000
>> >>>>> OutputStream
>> >>>>> 190
>> >>>>> 283
>> >>>>> 26
>> >>>>> 10
>> >>>>> 10,000
>> >>>>> OutputStream
>> >>>>> 392
>> >>>>> 542
>> >>>>> 25
>> >>>>> Summary: Under load Http Async Library approach is providing more throughput compared to Output stream approach.
>> >>>>> Both the approaches improve performance. The output stream approach can be used along with Http Async library approach which is giving around (3-5 ms) improvement in latency.
>> >>>>> Each approach is independent development. At this point I am keen to take up Http Async Library development.
>> >>>>> -- Thanks
>> >>>>> -- Kishore
>> >>>> --
>> >>>> Andrew Gaul
>> >>>> https://na01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fgaul.org%2F&data=02%7C01%7C%7C8690ee3e00bc4e3da0e608d4a406565e%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636313794511467457&sdata=XQ%2FshVjdqC3KiVEuyH6%2FJvmDN5DHBmS0kIBx98V89KY%3D&reserved=0
>>
>
>--
>Andrew Gaul
>https://na01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fgaul.org%2F&data=02%7C01%7C%7Cccc73352a8964b931b6a08d4b8d392a3%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636336666213699146&sdata=58ycimrkelgVaMn69lHY2nWE2qlzK6gpkuH%2FZP06M6E%3D&reserved=0
>
>
>
Re: Adding async support for Jclouds library
Posted by Battula Kishore <ba...@adobe.com.INVALID>.
Hi Andrew,
I can totally understand the concerns you have. Please let me know when you can start work on this? To reiterate this is one of the important feature for us and want to use it in our production systems. We wanted to contribute back the async changes to open source.
-- Thanks
-- Kishore
On 22/06/17, 12:00 AM, "Dancy, Chris" <Ch...@pega.com> wrote:
>Hey Everyone,
>
>I'm generally just a lurker on this list but we use jclouds internally here for various things. I've been following the async work and it sounds promising. If you need additional testing done please send around the details on how to do so and we'd be more than happy to help out and provide more benchmarks.
>
>Thanks,
>Chris
>________________________________________
>From: Andrew Gaul <ga...@apache.org>
>Sent: Wednesday, June 21, 2017 2:12 PM
>To: dev@jclouds.apache.org
>Subject: Re: Adding async support for Jclouds library
>
>As Andrew Philips said, I have little availability until early July.
>While others have capability of reviewing these changes, I prefer to
>look at them as well since we have screwed up a previous async API. I
>continue to doubt the async benchmark results and I want to further
>understand what you have measured. Given that we have been talking
>about these changes since March and earlier responses had no particular
>urgency, including ignoring my mails, we should not have any such
>urgency now. Assuming this is still open, I look forward to working on
>it in a few weeks.
>
>On Tue, Jun 06, 2017 at 06:28:06AM +0000, Felix Meschberger wrote:
>> Hi Jclouds dev
>>
>> Anyone else who could pick up instead of AndrewG ?
>>
>> It would be great to be able to have this support added to JClouds.
>>
>> Thanks
>> Felix
>>
>> > Am 06.06.2017 um 05:49 schrieb Andrew Phillips <an...@apache.org>:
>> >
>> > Hi Kishore
>> >
>> > Andrew G. is travelling for most of June, as far as I understand, so will likely be a bit slower to respond. Thanks for your patience!
>> >
>> > ap
>> >
>> > On 2017-06-05 23:42, Battula Kishore wrote:
>> >> Hi Andrew,
>> >> Any update on this?
>> >> -- Thanks
>> >> -- Kishore
>> >> On 29/05/17, 1:06 PM, "Battula Kishore" <ba...@adobe.com.INVALID> wrote:
>> >>> Hi Andrew,
>> >>> Thanks andrew for the quick response. Here is the GitHub repo and instructions on how to run the tests https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fkishore25kumar%2Fs3proxy-async-test-setup&data=02%7C01%7C%7C8969ed65c6a749aebfa608d4a66583cc%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636316402308126519&sdata=0YESeCSlmChUOicpe2m9SYC1WaB7m%2B7hIB6o%2BTOp1Jo%3D&reserved=0. The READE.md also has the repo details of s3proxy as well as jclouds implementation. Hope this helps? Let me know if you need anything else?
>> >>> In the mean time you review these results if you can let me know the design review process I can be prepared for that.
>> >>> -- Thanks
>> >>> -- Kishore
>> >>> On 26/05/17, 12:40 PM, "Andrew Gaul" <ga...@apache.org> wrote:
>> >>>> Kishore, these are promising results! I reformatted the most important
>> >>>> rows which show a 2x improvement in throughput and latency:
>> >>>> 10 10,000 Async Http Lib 209 282 48
>> >>>> 10 10,000 OutputStream 392 542 25
>> >>>> Can you share the implementation and include instructions on how to
>> >>>> replicate these tests?
>> >>>> On Wed, May 24, 2017 at 05:50:52AM +0000, Battula Kishore wrote:
>> >>>>> Hi,
>> >>>>> This is Kishore who is working on async poc using mail id(kishore25kumar@gmail.com<ma...@gmail.com>). I work at adobe and we wanted to implement async support for jclouds library and contribute it back.
>> >>>>> From the last discussion I was asked to get the performance numbers for the two approaches.
>> >>>>> Approach 1: Using Http Async Library
>> >>>>> Approach 2: Using Outputstream
>> >>>>> Test setup:
>> >>>>> 1. Both the s3 proxy server and test runner are running in same Docker container in azure west-us region.
>> >>>>> 2. Azure storage account is also residing in same west-us region.
>> >>>>> 3. A bucket is prepopulated with 100,000 files, each file of 1 MB size before test start.
>> >>>>> 4. The test runner sends unique requests to s3proxy to download files.
>> >>>>> Virtual Machine spec: CPU - 8 cores, Memory - 28 GB (Standard_D4 Azure machine)
>> >>>>> S3proxy is running with 1 jetty worker thread in all the scenarios. The payload size used is 1 MB file. Here are the performance numbers.
>> >>>>> Test Runner Threads
>> >>>>> Iteration Per thread
>> >>>>> Approach
>> >>>>> Avg response time (ms)
>> >>>>> 99%tile time (ms)
>> >>>>> Throughput
>> >>>>> (Requests / sec)
>> >>>>> 1
>> >>>>> 10,000
>> >>>>> Async Http Lib
>> >>>>> 45
>> >>>>> 87
>> >>>>> 22
>> >>>>> 5
>> >>>>> 10,000
>> >>>>> Async Http Lib
>> >>>>> 107
>> >>>>> 159
>> >>>>> 47
>> >>>>> 10
>> >>>>> 10,000
>> >>>>> Async Http Lib
>> >>>>> 209
>> >>>>> 282
>> >>>>> 48
>> >>>>> 1
>> >>>>> 10,000
>> >>>>> OutputStream
>> >>>>> 41
>> >>>>> 85
>> >>>>> 24
>> >>>>> 5
>> >>>>> 10,000
>> >>>>> OutputStream
>> >>>>> 190
>> >>>>> 283
>> >>>>> 26
>> >>>>> 10
>> >>>>> 10,000
>> >>>>> OutputStream
>> >>>>> 392
>> >>>>> 542
>> >>>>> 25
>> >>>>> Summary: Under load Http Async Library approach is providing more throughput compared to Output stream approach.
>> >>>>> Both the approaches improve performance. The output stream approach can be used along with Http Async library approach which is giving around (3-5 ms) improvement in latency.
>> >>>>> Each approach is independent development. At this point I am keen to take up Http Async Library development.
>> >>>>> -- Thanks
>> >>>>> -- Kishore
>> >>>> --
>> >>>> Andrew Gaul
>> >>>> https://na01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fgaul.org%2F&data=02%7C01%7C%7C8690ee3e00bc4e3da0e608d4a406565e%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636313794511467457&sdata=XQ%2FshVjdqC3KiVEuyH6%2FJvmDN5DHBmS0kIBx98V89KY%3D&reserved=0
>>
>
>--
>Andrew Gaul
>https://na01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fgaul.org%2F&data=02%7C01%7C%7Cccc73352a8964b931b6a08d4b8d392a3%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636336666213699146&sdata=58ycimrkelgVaMn69lHY2nWE2qlzK6gpkuH%2FZP06M6E%3D&reserved=0
>
>
>
Re: Adding async support for Jclouds library
Posted by "Dancy, Chris" <Ch...@pega.com>.
Hey Everyone,
I'm generally just a lurker on this list but we use jclouds internally here for various things. I've been following the async work and it sounds promising. If you need additional testing done please send around the details on how to do so and we'd be more than happy to help out and provide more benchmarks.
Thanks,
Chris
________________________________________
From: Andrew Gaul <ga...@apache.org>
Sent: Wednesday, June 21, 2017 2:12 PM
To: dev@jclouds.apache.org
Subject: Re: Adding async support for Jclouds library
As Andrew Philips said, I have little availability until early July.
While others have capability of reviewing these changes, I prefer to
look at them as well since we have screwed up a previous async API. I
continue to doubt the async benchmark results and I want to further
understand what you have measured. Given that we have been talking
about these changes since March and earlier responses had no particular
urgency, including ignoring my mails, we should not have any such
urgency now. Assuming this is still open, I look forward to working on
it in a few weeks.
On Tue, Jun 06, 2017 at 06:28:06AM +0000, Felix Meschberger wrote:
> Hi Jclouds dev
>
> Anyone else who could pick up instead of AndrewG ?
>
> It would be great to be able to have this support added to JClouds.
>
> Thanks
> Felix
>
> > Am 06.06.2017 um 05:49 schrieb Andrew Phillips <an...@apache.org>:
> >
> > Hi Kishore
> >
> > Andrew G. is travelling for most of June, as far as I understand, so will likely be a bit slower to respond. Thanks for your patience!
> >
> > ap
> >
> > On 2017-06-05 23:42, Battula Kishore wrote:
> >> Hi Andrew,
> >> Any update on this?
> >> -- Thanks
> >> -- Kishore
> >> On 29/05/17, 1:06 PM, "Battula Kishore" <ba...@adobe.com.INVALID> wrote:
> >>> Hi Andrew,
> >>> Thanks andrew for the quick response. Here is the GitHub repo and instructions on how to run the tests https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fkishore25kumar%2Fs3proxy-async-test-setup&data=02%7C01%7C%7C8969ed65c6a749aebfa608d4a66583cc%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636316402308126519&sdata=0YESeCSlmChUOicpe2m9SYC1WaB7m%2B7hIB6o%2BTOp1Jo%3D&reserved=0. The READE.md also has the repo details of s3proxy as well as jclouds implementation. Hope this helps? Let me know if you need anything else?
> >>> In the mean time you review these results if you can let me know the design review process I can be prepared for that.
> >>> -- Thanks
> >>> -- Kishore
> >>> On 26/05/17, 12:40 PM, "Andrew Gaul" <ga...@apache.org> wrote:
> >>>> Kishore, these are promising results! I reformatted the most important
> >>>> rows which show a 2x improvement in throughput and latency:
> >>>> 10 10,000 Async Http Lib 209 282 48
> >>>> 10 10,000 OutputStream 392 542 25
> >>>> Can you share the implementation and include instructions on how to
> >>>> replicate these tests?
> >>>> On Wed, May 24, 2017 at 05:50:52AM +0000, Battula Kishore wrote:
> >>>>> Hi,
> >>>>> This is Kishore who is working on async poc using mail id(kishore25kumar@gmail.com<ma...@gmail.com>). I work at adobe and we wanted to implement async support for jclouds library and contribute it back.
> >>>>> From the last discussion I was asked to get the performance numbers for the two approaches.
> >>>>> Approach 1: Using Http Async Library
> >>>>> Approach 2: Using Outputstream
> >>>>> Test setup:
> >>>>> 1. Both the s3 proxy server and test runner are running in same Docker container in azure west-us region.
> >>>>> 2. Azure storage account is also residing in same west-us region.
> >>>>> 3. A bucket is prepopulated with 100,000 files, each file of 1 MB size before test start.
> >>>>> 4. The test runner sends unique requests to s3proxy to download files.
> >>>>> Virtual Machine spec: CPU - 8 cores, Memory - 28 GB (Standard_D4 Azure machine)
> >>>>> S3proxy is running with 1 jetty worker thread in all the scenarios. The payload size used is 1 MB file. Here are the performance numbers.
> >>>>> Test Runner Threads
> >>>>> Iteration Per thread
> >>>>> Approach
> >>>>> Avg response time (ms)
> >>>>> 99%tile time (ms)
> >>>>> Throughput
> >>>>> (Requests / sec)
> >>>>> 1
> >>>>> 10,000
> >>>>> Async Http Lib
> >>>>> 45
> >>>>> 87
> >>>>> 22
> >>>>> 5
> >>>>> 10,000
> >>>>> Async Http Lib
> >>>>> 107
> >>>>> 159
> >>>>> 47
> >>>>> 10
> >>>>> 10,000
> >>>>> Async Http Lib
> >>>>> 209
> >>>>> 282
> >>>>> 48
> >>>>> 1
> >>>>> 10,000
> >>>>> OutputStream
> >>>>> 41
> >>>>> 85
> >>>>> 24
> >>>>> 5
> >>>>> 10,000
> >>>>> OutputStream
> >>>>> 190
> >>>>> 283
> >>>>> 26
> >>>>> 10
> >>>>> 10,000
> >>>>> OutputStream
> >>>>> 392
> >>>>> 542
> >>>>> 25
> >>>>> Summary: Under load Http Async Library approach is providing more throughput compared to Output stream approach.
> >>>>> Both the approaches improve performance. The output stream approach can be used along with Http Async library approach which is giving around (3-5 ms) improvement in latency.
> >>>>> Each approach is independent development. At this point I am keen to take up Http Async Library development.
> >>>>> -- Thanks
> >>>>> -- Kishore
> >>>> --
> >>>> Andrew Gaul
> >>>> https://na01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fgaul.org%2F&data=02%7C01%7C%7C8690ee3e00bc4e3da0e608d4a406565e%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636313794511467457&sdata=XQ%2FshVjdqC3KiVEuyH6%2FJvmDN5DHBmS0kIBx98V89KY%3D&reserved=0
>
--
Andrew Gaul
http://gaul.org/
Re: Adding async support for Jclouds library
Posted by Andrew Gaul <ga...@apache.org>.
As Andrew Philips said, I have little availability until early July.
While others have capability of reviewing these changes, I prefer to
look at them as well since we have screwed up a previous async API. I
continue to doubt the async benchmark results and I want to further
understand what you have measured. Given that we have been talking
about these changes since March and earlier responses had no particular
urgency, including ignoring my mails, we should not have any such
urgency now. Assuming this is still open, I look forward to working on
it in a few weeks.
On Tue, Jun 06, 2017 at 06:28:06AM +0000, Felix Meschberger wrote:
> Hi Jclouds dev
>
> Anyone else who could pick up instead of AndrewG ?
>
> It would be great to be able to have this support added to JClouds.
>
> Thanks
> Felix
>
> > Am 06.06.2017 um 05:49 schrieb Andrew Phillips <an...@apache.org>:
> >
> > Hi Kishore
> >
> > Andrew G. is travelling for most of June, as far as I understand, so will likely be a bit slower to respond. Thanks for your patience!
> >
> > ap
> >
> > On 2017-06-05 23:42, Battula Kishore wrote:
> >> Hi Andrew,
> >> Any update on this?
> >> -- Thanks
> >> -- Kishore
> >> On 29/05/17, 1:06 PM, "Battula Kishore" <ba...@adobe.com.INVALID> wrote:
> >>> Hi Andrew,
> >>> Thanks andrew for the quick response. Here is the GitHub repo and instructions on how to run the tests https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fkishore25kumar%2Fs3proxy-async-test-setup&data=02%7C01%7C%7C8969ed65c6a749aebfa608d4a66583cc%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636316402308126519&sdata=0YESeCSlmChUOicpe2m9SYC1WaB7m%2B7hIB6o%2BTOp1Jo%3D&reserved=0. The READE.md also has the repo details of s3proxy as well as jclouds implementation. Hope this helps? Let me know if you need anything else?
> >>> In the mean time you review these results if you can let me know the design review process I can be prepared for that.
> >>> -- Thanks
> >>> -- Kishore
> >>> On 26/05/17, 12:40 PM, "Andrew Gaul" <ga...@apache.org> wrote:
> >>>> Kishore, these are promising results! I reformatted the most important
> >>>> rows which show a 2x improvement in throughput and latency:
> >>>> 10 10,000 Async Http Lib 209 282 48
> >>>> 10 10,000 OutputStream 392 542 25
> >>>> Can you share the implementation and include instructions on how to
> >>>> replicate these tests?
> >>>> On Wed, May 24, 2017 at 05:50:52AM +0000, Battula Kishore wrote:
> >>>>> Hi,
> >>>>> This is Kishore who is working on async poc using mail id(kishore25kumar@gmail.com<ma...@gmail.com>). I work at adobe and we wanted to implement async support for jclouds library and contribute it back.
> >>>>> From the last discussion I was asked to get the performance numbers for the two approaches.
> >>>>> Approach 1: Using Http Async Library
> >>>>> Approach 2: Using Outputstream
> >>>>> Test setup:
> >>>>> 1. Both the s3 proxy server and test runner are running in same Docker container in azure west-us region.
> >>>>> 2. Azure storage account is also residing in same west-us region.
> >>>>> 3. A bucket is prepopulated with 100,000 files, each file of 1 MB size before test start.
> >>>>> 4. The test runner sends unique requests to s3proxy to download files.
> >>>>> Virtual Machine spec: CPU - 8 cores, Memory - 28 GB (Standard_D4 Azure machine)
> >>>>> S3proxy is running with 1 jetty worker thread in all the scenarios. The payload size used is 1 MB file. Here are the performance numbers.
> >>>>> Test Runner Threads
> >>>>> Iteration Per thread
> >>>>> Approach
> >>>>> Avg response time (ms)
> >>>>> 99%tile time (ms)
> >>>>> Throughput
> >>>>> (Requests / sec)
> >>>>> 1
> >>>>> 10,000
> >>>>> Async Http Lib
> >>>>> 45
> >>>>> 87
> >>>>> 22
> >>>>> 5
> >>>>> 10,000
> >>>>> Async Http Lib
> >>>>> 107
> >>>>> 159
> >>>>> 47
> >>>>> 10
> >>>>> 10,000
> >>>>> Async Http Lib
> >>>>> 209
> >>>>> 282
> >>>>> 48
> >>>>> 1
> >>>>> 10,000
> >>>>> OutputStream
> >>>>> 41
> >>>>> 85
> >>>>> 24
> >>>>> 5
> >>>>> 10,000
> >>>>> OutputStream
> >>>>> 190
> >>>>> 283
> >>>>> 26
> >>>>> 10
> >>>>> 10,000
> >>>>> OutputStream
> >>>>> 392
> >>>>> 542
> >>>>> 25
> >>>>> Summary: Under load Http Async Library approach is providing more throughput compared to Output stream approach.
> >>>>> Both the approaches improve performance. The output stream approach can be used along with Http Async library approach which is giving around (3-5 ms) improvement in latency.
> >>>>> Each approach is independent development. At this point I am keen to take up Http Async Library development.
> >>>>> -- Thanks
> >>>>> -- Kishore
> >>>> --
> >>>> Andrew Gaul
> >>>> https://na01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fgaul.org%2F&data=02%7C01%7C%7C8690ee3e00bc4e3da0e608d4a406565e%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636313794511467457&sdata=XQ%2FshVjdqC3KiVEuyH6%2FJvmDN5DHBmS0kIBx98V89KY%3D&reserved=0
>
--
Andrew Gaul
http://gaul.org/
Re: Adding async support for Jclouds library
Posted by Felix Meschberger <fm...@adobe.com.INVALID>.
Hi Jclouds dev
Anyone else who could pick up instead of AndrewG ?
It would be great to be able to have this support added to JClouds.
Thanks
Felix
> Am 06.06.2017 um 05:49 schrieb Andrew Phillips <an...@apache.org>:
>
> Hi Kishore
>
> Andrew G. is travelling for most of June, as far as I understand, so will likely be a bit slower to respond. Thanks for your patience!
>
> ap
>
> On 2017-06-05 23:42, Battula Kishore wrote:
>> Hi Andrew,
>> Any update on this?
>> -- Thanks
>> -- Kishore
>> On 29/05/17, 1:06 PM, "Battula Kishore" <ba...@adobe.com.INVALID> wrote:
>>> Hi Andrew,
>>> Thanks andrew for the quick response. Here is the GitHub repo and instructions on how to run the tests https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fkishore25kumar%2Fs3proxy-async-test-setup&data=02%7C01%7C%7C8969ed65c6a749aebfa608d4a66583cc%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636316402308126519&sdata=0YESeCSlmChUOicpe2m9SYC1WaB7m%2B7hIB6o%2BTOp1Jo%3D&reserved=0. The READE.md also has the repo details of s3proxy as well as jclouds implementation. Hope this helps? Let me know if you need anything else?
>>> In the mean time you review these results if you can let me know the design review process I can be prepared for that.
>>> -- Thanks
>>> -- Kishore
>>> On 26/05/17, 12:40 PM, "Andrew Gaul" <ga...@apache.org> wrote:
>>>> Kishore, these are promising results! I reformatted the most important
>>>> rows which show a 2x improvement in throughput and latency:
>>>> 10 10,000 Async Http Lib 209 282 48
>>>> 10 10,000 OutputStream 392 542 25
>>>> Can you share the implementation and include instructions on how to
>>>> replicate these tests?
>>>> On Wed, May 24, 2017 at 05:50:52AM +0000, Battula Kishore wrote:
>>>>> Hi,
>>>>> This is Kishore who is working on async poc using mail id(kishore25kumar@gmail.com<ma...@gmail.com>). I work at adobe and we wanted to implement async support for jclouds library and contribute it back.
>>>>> From the last discussion I was asked to get the performance numbers for the two approaches.
>>>>> Approach 1: Using Http Async Library
>>>>> Approach 2: Using Outputstream
>>>>> Test setup:
>>>>> 1. Both the s3 proxy server and test runner are running in same Docker container in azure west-us region.
>>>>> 2. Azure storage account is also residing in same west-us region.
>>>>> 3. A bucket is prepopulated with 100,000 files, each file of 1 MB size before test start.
>>>>> 4. The test runner sends unique requests to s3proxy to download files.
>>>>> Virtual Machine spec: CPU - 8 cores, Memory - 28 GB (Standard_D4 Azure machine)
>>>>> S3proxy is running with 1 jetty worker thread in all the scenarios. The payload size used is 1 MB file. Here are the performance numbers.
>>>>> Test Runner Threads
>>>>> Iteration Per thread
>>>>> Approach
>>>>> Avg response time (ms)
>>>>> 99%tile time (ms)
>>>>> Throughput
>>>>> (Requests / sec)
>>>>> 1
>>>>> 10,000
>>>>> Async Http Lib
>>>>> 45
>>>>> 87
>>>>> 22
>>>>> 5
>>>>> 10,000
>>>>> Async Http Lib
>>>>> 107
>>>>> 159
>>>>> 47
>>>>> 10
>>>>> 10,000
>>>>> Async Http Lib
>>>>> 209
>>>>> 282
>>>>> 48
>>>>> 1
>>>>> 10,000
>>>>> OutputStream
>>>>> 41
>>>>> 85
>>>>> 24
>>>>> 5
>>>>> 10,000
>>>>> OutputStream
>>>>> 190
>>>>> 283
>>>>> 26
>>>>> 10
>>>>> 10,000
>>>>> OutputStream
>>>>> 392
>>>>> 542
>>>>> 25
>>>>> Summary: Under load Http Async Library approach is providing more throughput compared to Output stream approach.
>>>>> Both the approaches improve performance. The output stream approach can be used along with Http Async library approach which is giving around (3-5 ms) improvement in latency.
>>>>> Each approach is independent development. At this point I am keen to take up Http Async Library development.
>>>>> -- Thanks
>>>>> -- Kishore
>>>> --
>>>> Andrew Gaul
>>>> https://na01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fgaul.org%2F&data=02%7C01%7C%7C8690ee3e00bc4e3da0e608d4a406565e%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636313794511467457&sdata=XQ%2FshVjdqC3KiVEuyH6%2FJvmDN5DHBmS0kIBx98V89KY%3D&reserved=0
Re: Adding async support for Jclouds library
Posted by Andrew Phillips <an...@apache.org>.
Hi Kishore
Andrew G. is travelling for most of June, as far as I understand, so
will likely be a bit slower to respond. Thanks for your patience!
ap
On 2017-06-05 23:42, Battula Kishore wrote:
> Hi Andrew,
>
> Any update on this?
>
> -- Thanks
> -- Kishore
>
>
>
>
>
>
>
>
>
>
> On 29/05/17, 1:06 PM, "Battula Kishore" <ba...@adobe.com.INVALID>
> wrote:
>
>> Hi Andrew,
>>
>> Thanks andrew for the quick response. Here is the GitHub repo and
>> instructions on how to run the tests
>> https://na01.safelinks.protection.outlook.com/?url=https%3A%2F%2Fgithub.com%2Fkishore25kumar%2Fs3proxy-async-test-setup&data=02%7C01%7C%7C8969ed65c6a749aebfa608d4a66583cc%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636316402308126519&sdata=0YESeCSlmChUOicpe2m9SYC1WaB7m%2B7hIB6o%2BTOp1Jo%3D&reserved=0.
>> The READE.md also has the repo details of s3proxy as well as jclouds
>> implementation. Hope this helps? Let me know if you need anything
>> else?
>>
>> In the mean time you review these results if you can let me know the
>> design review process I can be prepared for that.
>>
>> -- Thanks
>> -- Kishore
>>
>>
>>
>>
>>
>>
>>
>>
>>
>> On 26/05/17, 12:40 PM, "Andrew Gaul" <ga...@apache.org> wrote:
>>
>>> Kishore, these are promising results! I reformatted the most
>>> important
>>> rows which show a 2x improvement in throughput and latency:
>>>
>>> 10 10,000 Async Http Lib 209 282 48
>>> 10 10,000 OutputStream 392 542 25
>>>
>>> Can you share the implementation and include instructions on how to
>>> replicate these tests?
>>>
>>> On Wed, May 24, 2017 at 05:50:52AM +0000, Battula Kishore wrote:
>>>> Hi,
>>>>
>>>> This is Kishore who is working on async poc using mail
>>>> id(kishore25kumar@gmail.com<ma...@gmail.com>). I
>>>> work at adobe and we wanted to implement async support for jclouds
>>>> library and contribute it back.
>>>>
>>>> From the last discussion I was asked to get the performance numbers
>>>> for the two approaches.
>>>> Approach 1: Using Http Async Library
>>>> Approach 2: Using Outputstream
>>>>
>>>> Test setup:
>>>>
>>>> 1. Both the s3 proxy server and test runner are running in
>>>> same Docker container in azure west-us region.
>>>>
>>>> 2. Azure storage account is also residing in same west-us
>>>> region.
>>>>
>>>> 3. A bucket is prepopulated with 100,000 files, each file of 1
>>>> MB size before test start.
>>>>
>>>> 4. The test runner sends unique requests to s3proxy to
>>>> download files.
>>>>
>>>> Virtual Machine spec: CPU - 8 cores, Memory - 28 GB (Standard_D4
>>>> Azure machine)
>>>>
>>>> S3proxy is running with 1 jetty worker thread in all the scenarios.
>>>> The payload size used is 1 MB file. Here are the performance
>>>> numbers.
>>>> Test Runner Threads
>>>>
>>>> Iteration Per thread
>>>>
>>>> Approach
>>>>
>>>> Avg response time (ms)
>>>>
>>>> 99%tile time (ms)
>>>>
>>>> Throughput
>>>> (Requests / sec)
>>>>
>>>> 1
>>>>
>>>> 10,000
>>>>
>>>> Async Http Lib
>>>>
>>>> 45
>>>>
>>>> 87
>>>>
>>>> 22
>>>>
>>>> 5
>>>>
>>>> 10,000
>>>>
>>>> Async Http Lib
>>>>
>>>> 107
>>>>
>>>> 159
>>>>
>>>> 47
>>>>
>>>> 10
>>>>
>>>> 10,000
>>>>
>>>> Async Http Lib
>>>>
>>>> 209
>>>>
>>>> 282
>>>>
>>>> 48
>>>>
>>>> 1
>>>>
>>>> 10,000
>>>>
>>>> OutputStream
>>>>
>>>> 41
>>>>
>>>> 85
>>>>
>>>> 24
>>>>
>>>> 5
>>>>
>>>> 10,000
>>>>
>>>> OutputStream
>>>>
>>>> 190
>>>>
>>>> 283
>>>>
>>>> 26
>>>>
>>>> 10
>>>>
>>>> 10,000
>>>>
>>>> OutputStream
>>>>
>>>> 392
>>>>
>>>> 542
>>>>
>>>> 25
>>>>
>>>>
>>>> Summary: Under load Http Async Library approach is providing more
>>>> throughput compared to Output stream approach.
>>>>
>>>>
>>>> Both the approaches improve performance. The output stream approach
>>>> can be used along with Http Async library approach which is giving
>>>> around (3-5 ms) improvement in latency.
>>>>
>>>> Each approach is independent development. At this point I am keen to
>>>> take up Http Async Library development.
>>>>
>>>>
>>>> -- Thanks
>>>> -- Kishore
>>>>
>>>
>>> --
>>> Andrew Gaul
>>> https://na01.safelinks.protection.outlook.com/?url=http%3A%2F%2Fgaul.org%2F&data=02%7C01%7C%7C8690ee3e00bc4e3da0e608d4a406565e%7Cfa7b1b5a7b34438794aed2c178decee1%7C0%7C0%7C636313794511467457&sdata=XQ%2FshVjdqC3KiVEuyH6%2FJvmDN5DHBmS0kIBx98V89KY%3D&reserved=0