You are viewing a plain text version of this content. The canonical link for it is here.
Posted to user@hive.apache.org by Mario Casola <ma...@gmail.com> on 2013/06/14 17:54:13 UTC
hive to hbase mapping
Hi,
I have a performance issue when I query HBase from Hive.
My idea is to build the scenario below:
1. Collect data in hbase for aggregation purpose
2. Create an external table that map Hive to Hbase
3. Create a real Hive table
4. Periodically transfer data from hbase to Hive through "INSERTO INTO
<real hive table> SELECT * FROM <external table> WHERE time = 201305212909"
Currently I'm doing a test on a Hbase table that has 70,000,000 rows and
I'm trying to query this table with a single column value filter, like the
query above.
If I try this type of query directly in Hbase the response time is around
80 seconds.
If I try the query in Hive shell, after 30 minutes, all the tasks (9 in my
case) are 0,00% complete.
Which could be the problem?
thanks
Mario
Re: hive to hbase mapping
Posted by Mario Casola <ma...@gmail.com>.
Hi,
for the first question the answer is yes, with a 500,000 rows Hbase table,
the job complete successfully.
Second, the jobs are in running state. Attached you can see a syslog of one
of the jobs.
Third, I've tryed to set the "hbase.zookeeper.quorum" property but nothing
is changed.
Let me know if I can check other configurations.
thanks
Mario
2013/6/15 <ku...@gmail.com>
> Since your jobs are at 0%, it might actually be a problem with your hadoop
> cluster rather than hive. Couple of things to check would be:
>
> 1. Does a simple M/R job complete successfully?
> 2. Do logs for the jobs say something? Are the jobs in running state or
> pending state?
> 3. It is possible that job submitted from hive is unable to find the
> zookeeper quorum. To do that, you need to set "hbase.zookeeper.quorum"
> property in your hive-site.xml to point to your zookeeper quorum.
>
> Hope this helps.
>
> On Jun 14, 2013, at 11:54 AM, Mario Casola <ma...@gmail.com> wrote:
>
> Hi Sanjay,
>
> thanks for the response.
>
> I need Hbase because is perfect for aggregating data through the counters,
> and write performance is great.
> Now the problem is...Which is the best way for loading periodically (every
> hour for example) Hbase data in Hive table?
>
> Mario
>
>
>
> 2013/6/14 Sanjay Subramanian <Sa...@wizecommerce.com>
>
>> 6 months back I was tasked with building a Data platform for logs and I
>> benchmarked
>> Hbase + Hive (queries were 8X slower)
>> Hive only
>>
>> So I decided for Hive option and am deploying that solution to
>> production.
>>
>> Couple of things u can think while u design if u really want to go
>> HBase+Hive (also look at this
>> http://hadoopstack.com/hive-on-hbase-part-1/)
>> - Query only todays data in a Hive+Hbase architecture
>> - Older data than one day query Hive only
>>
>> Hope I am not diverting from your question and problem
>>
>> sanjay
>>
>> From: Mario Casola <ma...@gmail.com>
>> Reply-To: "user@hive.apache.org" <us...@hive.apache.org>
>> Date: Friday, June 14, 2013 8:54 AM
>> To: "user@hive.apache.org" <us...@hive.apache.org>
>> Subject: hive to hbase mapping
>>
>> Hi,
>>
>> I have a performance issue when I query HBase from Hive.
>> My idea is to build the scenario below:
>> 1. Collect data in hbase for aggregation purpose
>> 2. Create an external table that map Hive to Hbase
>> 3. Create a real Hive table
>> 4. Periodically transfer data from hbase to Hive through "INSERTO INTO
>> <real hive table> SELECT * FROM <external table> WHERE time = 201305212909"
>>
>> Currently I'm doing a test on a Hbase table that has 70,000,000 rows
>> and I'm trying to query this table with a single column value filter, like
>> the query above.
>> If I try this type of query directly in Hbase the response time is around
>> 80 seconds.
>> If I try the query in Hive shell, after 30 minutes, all the tasks (9 in
>> my case) are 0,00% complete.
>>
>> Which could be the problem?
>>
>> thanks
>> Mario
>>
>> CONFIDENTIALITY NOTICE
>> ======================
>> This email message and any attachments are for the exclusive use of the
>> intended recipient(s) and may contain confidential and privileged
>> information. Any unauthorized review, use, disclosure or distribution is
>> prohibited. If you are not the intended recipient, please contact the
>> sender by reply email and destroy all copies of the original message along
>> with any attachments, from your computer system. If you are the intended
>> recipient, please be advised that the content of this message is subject to
>> access, review and disclosure by the sender's Email System Administrator.
>>
>
>
Re: hive to hbase mapping
Posted by ku...@gmail.com.
Since your jobs are at 0%, it might actually be a problem with your hadoop cluster rather than hive. Couple of things to check would be:
1. Does a simple M/R job complete successfully?
2. Do logs for the jobs say something? Are the jobs in running state or pending state?
3. It is possible that job submitted from hive is unable to find the zookeeper quorum. To do that, you need to set "hbase.zookeeper.quorum" property in your hive-site.xml to point to your zookeeper quorum.
Hope this helps.
On Jun 14, 2013, at 11:54 AM, Mario Casola <ma...@gmail.com> wrote:
> Hi Sanjay,
>
> thanks for the response.
>
> I need Hbase because is perfect for aggregating data through the counters, and write performance is great.
> Now the problem is...Which is the best way for loading periodically (every hour for example) Hbase data in Hive table?
>
> Mario
>
>
>
> 2013/6/14 Sanjay Subramanian <Sa...@wizecommerce.com>
>> 6 months back I was tasked with building a Data platform for logs and I benchmarked
>> Hbase + Hive (queries were 8X slower)
>> Hive only
>>
>> So I decided for Hive option and am deploying that solution to production.
>>
>> Couple of things u can think while u design if u really want to go HBase+Hive (also look at this http://hadoopstack.com/hive-on-hbase-part-1/)
>> - Query only todays data in a Hive+Hbase architecture
>> - Older data than one day query Hive only
>>
>> Hope I am not diverting from your question and problem
>>
>> sanjay
>>
>> From: Mario Casola <ma...@gmail.com>
>> Reply-To: "user@hive.apache.org" <us...@hive.apache.org>
>> Date: Friday, June 14, 2013 8:54 AM
>> To: "user@hive.apache.org" <us...@hive.apache.org>
>> Subject: hive to hbase mapping
>>
>> Hi,
>>
>> I have a performance issue when I query HBase from Hive.
>> My idea is to build the scenario below:
>> 1. Collect data in hbase for aggregation purpose
>> 2. Create an external table that map Hive to Hbase
>> 3. Create a real Hive table
>> 4. Periodically transfer data from hbase to Hive through "INSERTO INTO <real hive table> SELECT * FROM <external table> WHERE time = 201305212909"
>>
>> Currently I'm doing a test on a Hbase table that has 70,000,000 rows and I'm trying to query this table with a single column value filter, like the query above.
>> If I try this type of query directly in Hbase the response time is around 80 seconds.
>> If I try the query in Hive shell, after 30 minutes, all the tasks (9 in my case) are 0,00% complete.
>>
>> Which could be the problem?
>>
>> thanks
>> Mario
>>
>> CONFIDENTIALITY NOTICE
>> ======================
>> This email message and any attachments are for the exclusive use of the intended recipient(s) and may contain confidential and privileged information. Any unauthorized review, use, disclosure or distribution is prohibited. If you are not the intended recipient, please contact the sender by reply email and destroy all copies of the original message along with any attachments, from your computer system. If you are the intended recipient, please be advised that the content of this message is subject to access, review and disclosure by the sender's Email System Administrator.
>
Re: hive to hbase mapping
Posted by Sanjay Subramanian <Sa...@wizecommerce.com>.
How about you have two streams - one to hbase and one to Hive fro your data generation source ?
Moving data out of Hbase may not be trivial specially if the data sizes are large….
From: Mario Casola <ma...@gmail.com>>
Reply-To: "user@hive.apache.org<ma...@hive.apache.org>" <us...@hive.apache.org>>
Date: Friday, June 14, 2013 9:54 AM
To: "user@hive.apache.org<ma...@hive.apache.org>" <us...@hive.apache.org>>
Subject: Re: hive to hbase mapping
Hi Sanjay,
thanks for the response.
I need Hbase because is perfect for aggregating data through the counters, and write performance is great.
Now the problem is...Which is the best way for loading periodically (every hour for example) Hbase data in Hive table?
Mario
2013/6/14 Sanjay Subramanian <Sa...@wizecommerce.com>>
6 months back I was tasked with building a Data platform for logs and I benchmarked
Hbase + Hive (queries were 8X slower)
Hive only
So I decided for Hive option and am deploying that solution to production.
Couple of things u can think while u design if u really want to go HBase+Hive (also look at this http://hadoopstack.com/hive-on-hbase-part-1/)
- Query only todays data in a Hive+Hbase architecture
- Older data than one day query Hive only
Hope I am not diverting from your question and problem
sanjay
From: Mario Casola <ma...@gmail.com>>
Reply-To: "user@hive.apache.org<ma...@hive.apache.org>" <us...@hive.apache.org>>
Date: Friday, June 14, 2013 8:54 AM
To: "user@hive.apache.org<ma...@hive.apache.org>" <us...@hive.apache.org>>
Subject: hive to hbase mapping
Hi,
I have a performance issue when I query HBase from Hive.
My idea is to build the scenario below:
1. Collect data in hbase for aggregation purpose
2. Create an external table that map Hive to Hbase
3. Create a real Hive table
4. Periodically transfer data from hbase to Hive through "INSERTO INTO <real hive table> SELECT * FROM <external table> WHERE time = 201305212909"
Currently I'm doing a test on a Hbase table that has 70,000,000 rows and I'm trying to query this table with a single column value filter, like the query above.
If I try this type of query directly in Hbase the response time is around 80 seconds.
If I try the query in Hive shell, after 30 minutes, all the tasks (9 in my case) are 0,00% complete.
Which could be the problem?
thanks
Mario
CONFIDENTIALITY NOTICE
======================
This email message and any attachments are for the exclusive use of the intended recipient(s) and may contain confidential and privileged information. Any unauthorized review, use, disclosure or distribution is prohibited. If you are not the intended recipient, please contact the sender by reply email and destroy all copies of the original message along with any attachments, from your computer system. If you are the intended recipient, please be advised that the content of this message is subject to access, review and disclosure by the sender's Email System Administrator.
CONFIDENTIALITY NOTICE
======================
This email message and any attachments are for the exclusive use of the intended recipient(s) and may contain confidential and privileged information. Any unauthorized review, use, disclosure or distribution is prohibited. If you are not the intended recipient, please contact the sender by reply email and destroy all copies of the original message along with any attachments, from your computer system. If you are the intended recipient, please be advised that the content of this message is subject to access, review and disclosure by the sender's Email System Administrator.
Re: hive to hbase mapping
Posted by Mario Casola <ma...@gmail.com>.
Hi Sanjay,
thanks for the response.
I need Hbase because is perfect for aggregating data through the counters,
and write performance is great.
Now the problem is...Which is the best way for loading periodically (every
hour for example) Hbase data in Hive table?
Mario
2013/6/14 Sanjay Subramanian <Sa...@wizecommerce.com>
> 6 months back I was tasked with building a Data platform for logs and I
> benchmarked
> Hbase + Hive (queries were 8X slower)
> Hive only
>
> So I decided for Hive option and am deploying that solution to
> production.
>
> Couple of things u can think while u design if u really want to go
> HBase+Hive (also look at this http://hadoopstack.com/hive-on-hbase-part-1/
> )
> - Query only todays data in a Hive+Hbase architecture
> - Older data than one day query Hive only
>
> Hope I am not diverting from your question and problem
>
> sanjay
>
> From: Mario Casola <ma...@gmail.com>
> Reply-To: "user@hive.apache.org" <us...@hive.apache.org>
> Date: Friday, June 14, 2013 8:54 AM
> To: "user@hive.apache.org" <us...@hive.apache.org>
> Subject: hive to hbase mapping
>
> Hi,
>
> I have a performance issue when I query HBase from Hive.
> My idea is to build the scenario below:
> 1. Collect data in hbase for aggregation purpose
> 2. Create an external table that map Hive to Hbase
> 3. Create a real Hive table
> 4. Periodically transfer data from hbase to Hive through "INSERTO INTO
> <real hive table> SELECT * FROM <external table> WHERE time = 201305212909"
>
> Currently I'm doing a test on a Hbase table that has 70,000,000 rows and
> I'm trying to query this table with a single column value filter, like the
> query above.
> If I try this type of query directly in Hbase the response time is around
> 80 seconds.
> If I try the query in Hive shell, after 30 minutes, all the tasks (9 in my
> case) are 0,00% complete.
>
> Which could be the problem?
>
> thanks
> Mario
>
> CONFIDENTIALITY NOTICE
> ======================
> This email message and any attachments are for the exclusive use of the
> intended recipient(s) and may contain confidential and privileged
> information. Any unauthorized review, use, disclosure or distribution is
> prohibited. If you are not the intended recipient, please contact the
> sender by reply email and destroy all copies of the original message along
> with any attachments, from your computer system. If you are the intended
> recipient, please be advised that the content of this message is subject to
> access, review and disclosure by the sender's Email System Administrator.
>
Re: hive to hbase mapping
Posted by Sanjay Subramanian <Sa...@wizecommerce.com>.
6 months back I was tasked with building a Data platform for logs and I benchmarked
Hbase + Hive (queries were 8X slower)
Hive only
So I decided for Hive option and am deploying that solution to production.
Couple of things u can think while u design if u really want to go HBase+Hive (also look at this http://hadoopstack.com/hive-on-hbase-part-1/)
- Query only todays data in a Hive+Hbase architecture
- Older data than one day query Hive only
Hope I am not diverting from your question and problem
sanjay
From: Mario Casola <ma...@gmail.com>>
Reply-To: "user@hive.apache.org<ma...@hive.apache.org>" <us...@hive.apache.org>>
Date: Friday, June 14, 2013 8:54 AM
To: "user@hive.apache.org<ma...@hive.apache.org>" <us...@hive.apache.org>>
Subject: hive to hbase mapping
Hi,
I have a performance issue when I query HBase from Hive.
My idea is to build the scenario below:
1. Collect data in hbase for aggregation purpose
2. Create an external table that map Hive to Hbase
3. Create a real Hive table
4. Periodically transfer data from hbase to Hive through "INSERTO INTO <real hive table> SELECT * FROM <external table> WHERE time = 201305212909"
Currently I'm doing a test on a Hbase table that has 70,000,000 rows and I'm trying to query this table with a single column value filter, like the query above.
If I try this type of query directly in Hbase the response time is around 80 seconds.
If I try the query in Hive shell, after 30 minutes, all the tasks (9 in my case) are 0,00% complete.
Which could be the problem?
thanks
Mario
CONFIDENTIALITY NOTICE
======================
This email message and any attachments are for the exclusive use of the intended recipient(s) and may contain confidential and privileged information. Any unauthorized review, use, disclosure or distribution is prohibited. If you are not the intended recipient, please contact the sender by reply email and destroy all copies of the original message along with any attachments, from your computer system. If you are the intended recipient, please be advised that the content of this message is subject to access, review and disclosure by the sender's Email System Administrator.