[OPEN-ILS-DEV] Evergreen & Software Performance Analysis

[Kathy Lussier]

Hi all,

It was brought to my attention that everyone who may have some input for 
this discussion may not have an Evergreen wiki account. In that case, 
please feel free to send an e-mail (to the list, not directly to me) 
identifying any performance issues you believe should be addressed 
through a performance evaluation. I'll be happy to add them to the wiki.

What I'm looking for is:

1. Any specific paint points you see in performance.
2. Any specific questions you think a performance evaluation should answer.
3. Any ideas you might already have regarding causes of performance 
problems. In reading through the logs from the "future of the staff 
client" meeting, I noticed several people said they thought it was 
important to bring these ideas together before reaching out to a 
consultant, and I agree that this is an important first step in the process.

I posted just a few of our local issues at 
http://www.evergreen-ils.org/dokuwiki/doku.php?id=dev:testing:performance_issues:

STAFF CLIENT:

Memory leaks - is there an inherent problem with the technology used in 
the staff client (xulrunner, Dojo) that is the source of the memory leak 
problem and other performance problems?
Slow retrieval of patron records

MESSAGING (OPENSRF):
Staff client batch operations (e.g. updates/deletes from copy buckets)

DATABASE:
Catalog search - is there a way to optimize searching in the catalog so 
that users get faster results and are able to start re-implementing 
things like search.relevance_adjustment to provide boosts to relevance 
ranking?

I'm quite sure there are far more pain points out there, so please don't 
feel shy about contributing to the list!

Kathy



Kathy Lussier
Project Coordinator
Massachusetts Library Network Cooperative
(508) 343-0128
klussier at masslnc.org
Twitter: http://www.twitter.com/kmlussier

On 2/25/2013 11:44 AM, Kathy Lussier wrote:
> Hi all,
>
> Having heard no objections to proceeding with finding somebody to do a 
> software performance analysis, I have created a page on the wiki at 
> http://www.open-ils.org/dokuwiki/doku.php?id=dev:testing:performance_issues 
> where we can identify the pain points that need further evaluation and 
> add any questions that we hope a performance analysis might be able to 
> answer.
>
> I have started the list off with some basic issues/questions that have 
> come up in our own systems. During the future of the staff client 
> meeting, Dan Scott had mentioned that there might be three points of 
> attack:client, opensrf, database.  I thought dividing the list into 
> those three areas might be a good way to start.
>
> I'm hoping that all the knowledgeable sys admins out there who have a 
> stronger understanding of the system architecture than I do can build 
> this list into something that might be a good starting point for any 
> performance evaluation, whether it's done by a third party or by 
> somebody in the Evergreen community. By identifying the questions we 
> hope a performance evaluation might answer, we are also identifying 
> what our expectations are before we enter the process. I would want to 
> be clear on our expectations before formally talking to any third 
> party so that we can be fully informed about whether an evaluation 
> could meet those expectations.
>
> Kathy
>
>
> Kathy Lussier
> Project Coordinator
> Massachusetts Library Network Cooperative
> (508) 343-0128
> klussier at masslnc.org
> Twitter:http://www.twitter.com/kmlussier
> On 2/20/2013 2:26 PM, Mike Rylander wrote:
>> On Wed, Feb 20, 2013 at 2:10 PM, Kathy Lussier <klussier at masslnc.org 
>> <mailto:klussier at masslnc.org>> wrote:
>>
>>     Hi all,
>>
>>     I wasn't sure if I should add this to the QA discussion, but it
>>     seemed worthy of its own thread.
>>
>>     During the "future of the staff client" meeting, I advocated for
>>     bringing in a consultant to do a software performance analysis
>>     for Evergreen to help us identify where the critical bottlenecks
>>     are in the system in the hopes that we could then identify the
>>     areas that need to be worked on to improve performance. At the
>>     time, I didn't have any concrete suggestions on finding a
>>     consultant who could take on this project, but I have since done
>>     some more investigation and have a couple of leads, the most
>>     promising of which is an individual local to Massachusetts who
>>     previously worked for many years at Stratus Technologies where he
>>     was involved in all levels of performance analysis. He now
>>     teaches graduate-level courses on performance evaluation and also
>>     does contract work.
>>
>>     Now that I actually have concrete leads, I would like to get the
>>     ball rolling, provided there is support from the larger
>>     community. I'm not quite sure how this might fit in with ESI's
>>     planned QA efforts or with the possibility of bringing in a firm
>>     like OmniTI as Dan suggested, but my reading into these QA
>>     e-mails is that the focus would be on testing new commits.
>>
>>
>> I want to clarify something that Dan seems to have assumed 
>> incorrectly: that anything ESI does is mutually exclusive with 
>> bringing in outside expertise.  Nobody has any grounds to stop such 
>> an effort, and it would be ridiculous to argue otherwise, words put 
>> into my mouth notwithstanding.  The initial focus of an ESI effort 
>> will be what exists today, through infrastructure, so that what 
>> exists tomorrow can then be tested.
>>
>> As for how it would fit in, ESI would absorb and internalize any 
>> advice or direction, just like any other community member, and work 
>> within the community to incorporate that.
>>
>> So, why have ESI involved at all?  Besides the fact that we create a 
>> significant portion of the code, and that it benefits us as much as 
>> anyone to have a more stable Evergreen, there is a need for ongoing, 
>> active leadership in QA.  The fact is that it has not materialized 
>> yet, so we're looking for a way to make that a maintainable 
>> proposition for the community's benefit.  That means ongoing, deep 
>> integration with both developer and user communities.  And that is 
>> not something that we can expect from OmniTI or any other 
>> organization that is not plugged into those communities.  Could some 
>> other organization step into that role, and provide years of ongoing 
>> QA support?  Perhaps so, but ESI exists today and has the Evergreen 
>> expertise needed to avoid long (and costly) ramp-up time.
>>
>> The point is this, though, ESI will encourage any effort to improve 
>> Evergreen, and is willing and able to work in the community, as we 
>> always do, to further those efforts.
>> Thanks, Kathy!
>>
>> -- 
>> Mike Rylander
>>  | Director of Research and Development
>>  | Equinox Software, Inc. / Your Library's Guide to Open Source
>>  | phone:  1-877-OPEN-ILS (673-6457)
>>  | email: miker at esilibrary.com <mailto:miker at esilibrary.com>
>>  | web: http://www.esilibrary.com
>

-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://libmail.georgialibraries.org/pipermail/open-ils-dev/attachments/20130227/f3c3a076/attachment.htm>