Srinath, Srinath Perera wrote: > Hi Eranda; > > If you need more than average, right way to go is drawing confidence > intervals (http://en.wikipedia.org/wiki/Confidence_interval)/ error > bars, which will let you draw both servers on same graph. we already planned to implement this in the next step. I will add a wiki with the TODO list. > But in > general, only average is good. > > Also what are X/Y axis? > If you look at the detailed graphs, X-axis represent the demand rate, which is the number requests sent per second. For example starting from 100, increasing by steps of 50 up to 500. Y axis represent the replies per second. Graphs are for minimum reply rate, maximum reply rate, average reply rate and standard deviation.
There are so many graphs. Eranda has sent graphs related to reply rate only. Can someone suggest a good way to represent all the graphs so that one can grasp the results easily? Thanks, Damitha > Thanks > Srinath > > > > On Tue, Jun 1, 2010 at 5:16 PM, Eranda Sooriyabandara <[email protected]> wrote: > >> This test was done for the Data Services team to compare the performance of >> the Data Services servers 2.2.0 and 2.5.0 using the Performance testing >> framework. >> What I did here is add a RDBMS data source, create a data query service, and >> querying different number of rows form the Data Service server. >> In performance test framework we can increase the number of concurrent >> requests per second(demand rate) step by step so that we can exactly get the >> point where server begin to saturate. I tested for 100, 200, .... 1000 >> query rows for 100, 120, 140, .... ,500 demand rates and get the results. >> Here I am attaching the results with this. In results.png, include a quick >> review of all results and in the zip file there are separate graphs in >> detail. >> >> thoughts: >> Data Services 2.5.0 shows slightly high performance(almost the same) than >> Data Service 2.2.0. >> We can identify the demand rate where the server start to saturate, which >> shows the highest response rate. When query rows increases, it saturate at >> lower demand rates. >> >> Ideally this test is to be done in separate client and server machines. But >> I ran them in same machine. Shall I use the private cloud to do the testing >> for separate client/server? >> -- >> Thanks >> Eranda >> >> _______________________________________________ >> Carbon-dev mailing list >> [email protected] >> https://mail.wso2.org/cgi-bin/mailman/listinfo/carbon-dev >> >> >> > > > > -- __________________________________________________________________ Damitha Kumarage Technical Lead; WSO2 Inc. "Oxygenating the Web Service Platform; " http://www.wso2.com/ blog: " http://damithakumarage.wordpress.com/ __________________________________________________________________ _______________________________________________ Carbon-dev mailing list [email protected] https://mail.wso2.org/cgi-bin/mailman/listinfo/carbon-dev
