Is WLM usable only in a Virtualized environment?

[PrasannaKumari]

There is a requirement to provide the maximum CPU and the memory utilizations of one of our agents which is a java application on HP-UX. The intention is to stop our agent when it is trying to use more CPU or memory on a HP-UX host than these values. This is because there are other critical applications running on the HP-UX host and these applications have strict SLAs w.r.t CPU and memory. So the objective is to make sure that our agent is stopped so that these critical applications are not affected.

Finding the max CPU and memory utilization is a variant of the host configuration as per our agent's behaviour and hence will vary with different configurations of the host.

Can WLM be used for solving this problem? According to the documentation it appears that WLM is used only on top a virtualization solution of HP-UX. If so, is there any other way of handling this case?

[Duncan Edmonstone]

Yes WLM (or gWLM) can be used for this sort of thing - WLM is basically a control engine that drives a number of resource management functions - most of those resource management functions are partitioning technologies such as vPars or HPVMs, but WLM can also be used to control PRM or PSETs, so that will allow you to do resource management within a single OS image.

Based on your problem statement above, it seems that your requirements are fairly static, so you might not even need WLM. You may find that PRM will do everyting you need. With PRM its pretty simple to assign CPU/memory resources to processes and limit their usage when there is resource contention. I'd start with this manual here:

http://docs.hp.com/en/B8733-90027/PRM.ug.pdf

HTH

Duncan