<div dir="ltr">It's interesting and practical!<div>There're some early efforts to make openstack more robust and efficient, however, we're still lacking one such framework.</div><div><br></div><div>Just a little question.<br>
<div style>On the wiki, it says to consider the scalability problem.</div><div style>How can we measure the performance at large scale? By real deployment or just simulating the input?</div><div style>Through the descriptions, seems the method is to try testing every single component.</div>
<div style><br></div><div style>Also, suggest adding necessary <span id="213f2e81-1548-4293-8509-a811cb0ef32d" class="GINGER_SOFTWARE_mark">perf</span> checks into the test/integration jobs, too.</div><div style>Thanks!</div>
<div style><br></div></div></div><div class="gmail_extra"><br><br><div class="gmail_quote">On Tue, Jul 22, 2014 at 5:53 AM, Boris Pavlovic <span dir="ltr"><<a href="mailto:boris@pavlovic.me" target="_blank">boris@pavlovic.me</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="margin:0 0 0 .8ex;border-left:1px #ccc solid;padding-left:1ex"><div dir="ltr">Hi Stackers and TC,<br><br>The Rally contributor team would like to propose a new OpenStack program<br>
with a mission to provide scalability and performance benchmarking, and<br>
code profiling tools for OpenStack components.<br>
<br>We feel we've achieved a critical mass in the Rally project, with an<br>active, diverse contributor team. The Rally project will be the initial<br>project in a new proposed "Performance and Scalability" program.<br>
<br>Below, the details on our proposed new program.<br><br>Thanks for your consideration,<br>Boris<div><br></div><div><br><br>[1] <a href="https://review.openstack.org/#/c/108502/" target="_blank">https://review.openstack.org/#/c/108502/</a><div>
<br><div><div><br></div><div>Official Name<br>=============<br><br>Performance and Scalability<br><br>Codename<br>========<br><br>Rally<br><br>Scope<br>=====<br><br>Scalability benchmarking, performance analysis, and profiling of<br>
OpenStack components and workloads<br><br>Mission<br>=======<br><br>To increase the scalability and performance of OpenStack clouds by:<br><br>* defining standard benchmarks<br>* sharing performance data between operators and developers<br>
* providing transparency of code paths through profiling tools<br><br>Maturity<br>========<br><br>* Meeting logs <a href="http://eavesdrop.openstack.org/meetings/rally/2014/" target="_blank">http://eavesdrop.openstack.org/meetings/rally/2014/</a><br>
* IRC channel: #openstack-rally<br>* Rally performance jobs are in (Cinder, Glance, Keystone & Neutron)<br>check pipelines.<br>* > 950 commits over last 10 months<br>* Large, diverse contributor community<br> * <a href="http://stackalytics.com/?release=juno&metric=commits&project_type=All&module=rally" target="_blank">http://stackalytics.com/?release=juno&metric=commits&project_type=All&module=rally</a></div>
<div> * <a href="http://stackalytics.com/report/contribution/rally/180" target="_blank">http://stackalytics.com/report/contribution/rally/180</a><br><br>* Non official lead of project is Boris Pavlovic<br> * Official election In progress.<br>
<br>Deliverables<br>============<br><br>Critical deliverables in the Juno cycle are:<br><br>* extending Rally Benchmark framework to cover all use cases that are<br>required by all OpenStack projects<br>* integrating OSprofiler in all core projects<br>
* increasing functional & unit testing coverage of Rally.<br><br>Discussion<br>==========<br><br>One of the major goals of Rally is to make it simple to share results of<br>standardized benchmarks and experiments between operators and<br>
developers. When an operator needs to verify certain performance<br>indicators meet some service level agreement, he will be able to run<br>benchmarks (from Rally) and share with the developer community the<br>results along with his OpenStack configuration. These benchmark results<br>
will assist developers in diagnosing particular performance and<br>scalability problems experienced with the operator's configuration.<br><br>Another interesting area is Rally & the OpenStack CI process. Currently,<br>
working on performance issues upstream tends to be a more social than<br>technical process. We can use Rally in the upstream gates to identify<br>performance regressions and measure improvement in scalability over<br>time. The use of Rally in the upstream gates will allow a more rigorous,<br>
scientific approach to performance analysis. In the case of an<br>integrated OSprofiler, it will be possible to get detailed information<br>about API call flows (e.g. duration of API calls in different services).</div></div>
</div><div><br></div><div><br></div></div></div>
<br>_______________________________________________<br>
OpenStack-dev mailing list<br>
<a href="mailto:OpenStack-dev@lists.openstack.org">OpenStack-dev@lists.openstack.org</a><br>
<a href="http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev" target="_blank">http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev</a><br>
<br></blockquote></div><br><br clear="all"><div><br></div>-- <br><font color="#999999">Best wishes!<br>Baohua<br></font>
</div>