cancel
Showing results for 
Search instead for 
Did you mean: 

Internet Sales Application Performance

Former Member
0 Kudos

We currently have a b2c site and two b2b sites deployed on CRM 3.0 ISA. The total number of concurrent users could be 50 - 100 with around 700 transactions created daily between the three sites.

We are struggling with reported performance issues and have tried to come up with what might be some benchmarks for certain ISA transactions from the end user experience. Our web end user monitoring services can tell us how many seconds a user will wait for certain transactions and the reports we get vary from 2 - 10 seconds per click on the b2b sites for example.

Does SAP publish any performance benchmarks in regards to end user response times for their ISA applications? Has anyone else struggled in the response time area around their ISA applications?

Also in regards to landscapes, we have 2 application servers and one J2EE server to handle the load mentioned in my first paragraph. Does that seem reasonable?

I know this is a very general question but didn't know if anyone else has come across these issues. Any insight would be appreciated.

Thanks,

Will

Accepted Solutions (1)

Accepted Solutions (1)

former_member197561
Active Participant
0 Kudos

Dear Wili,

The ISA is application which consists of Frontend (browser), Java server and ABAP server parts. The most important  for the analysis is to determine how is the response time distributed between the different layers:

1) use /nSTAD to check response time in ABAP backend

2) use jDSR (in Netweaver WebAdministrator, or direct URL http://<host>:50000/dsr which redirects to http://<host>:50013/rsd/JavaStatistics.html ... , also consider the correct htp port for your instance ) to determine response time of Java server

3) use HttpWatch to determine the time at the browser side

If the dominant part of response time is in ABAP server, check DB time and CPU time. If DB time is high - proceed analysis by using /nST05, if CPU time is high, proceed analysis with /nSE30.

If dominant time is on java server side, proceed analysis with SAP JVM Profiler

http://www.sdn.sap.com/irj/scn/downloads?rid=/library/uuid/a0a67d7d-36e9-2a10-3a99-94fc8caaea0e

Hotspot snapshot and Memory allocation snapshot ; Do not forget to check the GC activities  and JVM settings.

If the dominant time is delta between browser time and server time, then check the network or the virus scan software installed at the frontend machine of the end user.

Best Regards,

Sylvia

Answers (2)

Answers (2)

Former Member
0 Kudos

Hi Will ,

Do you remember how did you tackel the above performance problem Any suggestion would be really appreciated .

Thanks and Regards,

Ankur

Former Member
0 Kudos

Will,

Did you come up with any measures of solving this performance issue? We are also stuck up in the same kind of scenario.

Any help would be invaluable.