Short answer, at least 1, up to the limit of the resources or address space
on the host.
James Pulley, http://www.loadrunnerbythehour.com/PricingMatrix
hi All,
thanks
Jayaprakash
--
You received this message because you are subscribed to the Google
"LoadRunner" group.
To post to this group, send email to LR-Loa...@googlegroups.com
To unsubscribe from this group, send email to
LR-LoadRunne...@googlegroups.com
For more options, visit this group at
http://groups.google.com/group/LR-LoadRunner?hl=en
It is a bit more complex than just a RAM calculation. The performance footprint that we consider in the analysis stage consists of being bounded on Memory, Disk, Network or CPU. Your virtual users, depending upon protocol and how they are constructed could become bound on any one (or multiple) of the above resource areas. Here is an opportunity to exercise your analytical skills for bottleneck analysis by monitoring your load generators and applying the same processes for bottleneck identification to your load generators as you apply to your application under test.
Your friend also appears to have a process issue by using the controller as a Load Generator, in addition to the control console for the test. The front end action of the LoadRunner controller is best served by residing on a dedicated computer so the video demands of the controller activity does not distort the performance envelope/resource window for lower priority application processes, such as virtual users. On your load generators go for blank screen savers and the lightest OS install you can manage in your environment. If you log information, go for a co-processed disk subsystem which takes both IRQ and INT load off of your core system. If you have a lot of Network Activity, consider the use of an intelligent network adapter with a TCPOE (TCP Offload Engine) to once again take the IRQ and INT load off of your core system.
Load Generation will get a lot easier with LR 11 (Just released). With 64 bit load generation we now have lifted the limit on RAM resources on the Load Generator beyond the 2GB (Windows Workstation) and 4GB (Windows Server) to 2^64, which hopefully we will not hit for a long time. With the addressable RAM limit addressed we’ll very likely see the other bottleneck areas come to the forefront, DISK and NETWORK on the generator. Sort of like squeezing one end of a balloon and observing the air bulge out the other end.
James Pulley, http://www.loadrunnerbythehour.com/PricingMatrix