Re: [GSoC - 2015] [Project Deliverables]

[Imesh Gunaratne <im...@apache.org>]

Hi Pranavan,

It's nice to hear from you! Yes I went through the attached documents,
please find few comments below:

   - Documentation has only around 3 pages of content, might be better to
   explain this in detail
   - The bench mark test has been run for around 20 min and it has only
   done once. IMO this is not sufficient to take a decision on the autoscaling
   improvements we have done.
   - The parameters of the bench mark test have not been mentioned; IaaS,
   cartridge, instance type, load average, memory, request load, etc.
   - It would be better to draw a graph to illustrate the actual instance
   count needed and the ones spawned. This would clearly show us how accurate
   autoscaling decisions were.
   - Can you please explain what exactly "Old Value" and "New Value" are in
   the bench mark test results spreadsheet?

Thanks

On Sun, Nov 22, 2015 at 4:56 PM, Pranavan Theivendiram <
pranavan.11@cse.mrt.ac.lk> wrote:

> Hi Imesh,
>
> Did you look over these docs?
>
> Thanks
>
> *T. Pranavan*
> *BSc Eng Undergraduate| Department of Computer Science & Engineering
> ,University of Moratuwa*
> *Mobile| *0775136836
>
> On 27 August 2015 at 00:52, Pranavan Theivendiram <
> pranavan.11@cse.mrt.ac.lk> wrote:
>
>> Hi All,
>>
>> I have done a final demo today. Please find the Project Deliverables
>> below.
>>
>> 1. New Pull Request - [1] <https://github.com/apache/stratos/pull/451>
>> 2. Presentation - [2]
>> <https://docs.google.com/presentation/d/1pCsvTxorHr20pCLeTgs2ft8kz8ckMcD3fzs7f3BQN2o/edit?usp=sharing>
>> 3. Demo Screen cast - [3] <https://www.youtube.com/watch?v=oSt3lqu2E7Y>
>> 4. Bench Mark Test - [4]
>> <https://docs.google.com/spreadsheets/d/1PRuTgc1LxRMcb21p-ZWBwX_JaMqVaVxixTZ4__0x3Pk/edit?usp=sharing>
>> 5. Patch file needed to use with CEP 3.1- [5]
>> <https://drive.google.com/a/cse.mrt.ac.lk/file/d/0B2WKC4wY2HQgUXBqYjFTRzJEcUE/view?usp=sharing>
>>
>> As Lahiru suggested, I am improving docs[6]
>> <https://docs.google.com/document/d/1X-9rluxB_CCXbqFH7Zd5xkgCAnkt2uTE2FzVvM3rJi4/edit?usp=sharing>
>> by adding the logic that I have used.
>>
>> Two main points have discussed for the future improvement.
>> 1. Finding the optimal number of points where the predictor value is more
>> accurate
>>  2. Try the approach with 3rd order polynomials
>>
>> Thanks
>> *T. Pranavan*
>> *BSc Eng Undergraduate| Department of Computer Science & Engineering
>> ,University of Moratuwa*
>> *Mobile| *0775136836
>>
>
>


-- 
Imesh Gunaratne

Senior Technical Lead, WSO2
Committer & PMC Member, Apache Stratos