
As we now have a CRM 7.0 EhP 1 development system my first try was to test if there are measurable performance improvements. And the good new is: They are really big improvements.
To get reproducible and exact test results I've utilized my {code:html}SAP Code Exchange{code} project SeleniumABAP (https://cw.sdn.sap.com/cw/groups/seleniumabap). First I've recorded the test steps using Selenium IDE (http://seleniumhq.org/projects/ide/). The saved test is then converted into a ABAP Unit test. To measure the traffic in bytes and the execution time in seconds I've extended this test with calls of the SeleniumABAP Client method capturenetworktraffic and the ABAP call GET RUN TIME FIELD.
The test was executed on a computer with a Core i5 650 CPU, 3.2 GHz, 2 GB RAM, Windows XP, 32 bit running Firefox 3.6.13 and using Selenium Server Standalone version 2.0b1 (http://code.google.com/p/selenium/downloads/list). The computer has a 1 GBit/s LAN connection to the CRM Servers.The 7.0 and 7.0 EhP 1 system contains around 160,000 Business Partners in table BUT000. The 2007 (6.0) system is the IDES System containing 11,500 Business Partners.
The most impressive improvement is the reduction of the initial size to start the CRM WebUI Framework. In CRM 7.0 that was 1,486,533 bytes. Now with the EhP 1 it was reduced to 563,992. Thats a reduction of 62 %. Well done CRM WebUI team! Unfortunately with a filled browser cache EhP 1 increased it's traffic footprint by 22 % against 7.0 and by 48 % compared to 2007 (6.0). Please note that the scale for the empty browser cache measurement starts at 500,000 Bytes to show also the little impact the which the other pages have. Also the graph for the filled browser cache does not start at zero.
The improvements on the execution time are not so big as for the traffic but also noticeable. Against CRM 7.0 32 % with an empty browser cache and 27 % with a filled cache. Please note that the graph starts at 2 seconds to show more details.
You must be a registered user to add a comment. If you've already registered, sign in. Otherwise, register and sign in.