1) Regarding “2 instances”, do you mean to separate Business Objects environments? If so is this possible with XI R2? I don’t recall, its been a while.

2) What is your expected peak concurrent Web Intelligence use level? In other words, how many WebI users do you expect at any peak time?

3) Is this server running anything else?

I was wondering if you or anyone else has recommendations on how many WebI Servers and connections per Webi Server is best for the following on R2:

4 quad core CPU’s
12 cores total (we disabled 4 cores because of licensing)
64RAM
2 instances

I appreciate your help.

I’ve done benchmark tests between XI 3.1 SP3 and XI R2 thanks to JMeter.
Because I want to upgrade my BO.

And the results are alarming.
I worked with two Windows platform (one R2, one 3.1) that have the same physical configuration :
4CPU, 1CMS, 1WebIReportServer

Let’s focus on the following scenario :
Login
Loop 15 times => Open a report
Logoff

Test properties :
Ramp-Up Period (in seconds) = 150
Duration = 600
LoopCount = Forever

Results Average (seconds)
For 1 thread :
Login R2 = 0,260
Login 3.1 = 2,731
OPEN R2 = 1,014
OPEN 3.1 = 2,061
Logoff R2 = 0,127
Logoff 3.1 = 0,363

For 20 threads :
Login R2 = 0,146
Login 3.1 = 2,702
OPEN R2 = 1,101
OPEN 3.1 = 5,245
Logoff R2 = 0,107
Logoff 3.1 = 0,498

For 50 threads :
Login R2 = 0,266
Login 3.1 = 2,681
OPEN R2 = 3,720
OPEN 3.1 = 14,179
Logoff R2 = 0,227
Logoff 3.1 = 0,563

Are you surprised by that ?
Or can you confirm these figures by your experience ?
The 3.1 WIRS is slower than R2 WIRS ??

A server with 4 core (1 quad core CPU, or 2 dual-core CPUs) could probably handle this, but having 6 or more cores would be better. 4 GB of RAM would be good. You will need at least 40 GB of storage, but more would be nice for times when extra logging is needed. Of course, it would be better if you could have two boxes and have each house at least 1 of each kind of server (for high-availability).

Your article is really interesting & seems very useful. Let me just mention my query here, I have my client with total 75 USERS out of them 25 would be Advanced users with Ad-Hoc data analysis capabilities & rest 50 would just be responsible to view those designed reports. All these 75 users would be concurrently accessing the server & will also include designing of reports & refresh.

I would really appreciate if you could kindly provide me insights on server sizing.

Awaiting response,
Chaitanya

Thank you for quick reply,
Currently we have maximum jobs set to 15.
we usually monitor them through task manager as you specified but what i guess is they seem to be dead inside and queued up.
so to overcome this situation we would like to set up Probes to poll on specific criteria.

Just wanted to know if any one has done monitoring other than through task manager?

Monitoring is a very different topic. The best you can do is to monitor the executables for the job server and the 3 WebI server. Assuming you are running on Windows then you would merely need to login to the server and look at Task Manager to get the exact process names. It would be best if you could monitor that all 3 WIRS and the 1 job server are always up. If you wanted to monitor functionality then you would have to find a way to combine your jobs with an event or program job that would write out a file for your monitor to check on. I am interested to see if anyone else has any recommendations on monitoring WebI and Job servers.

We have scheduled WebI reports that are pending in queue and also number of jobs running in queue are low compared to jobs pending in queue.

We are planning to monitor the processes and use Probes. Can you please specify which processes to be monitiored for scheduled webI reports?

Also can you please provide me how to design the requiremnts for monitoring the processes?

Currently we have 2 CPU’s and 3 WIRS.

Regards,
Nickie