Getting freequent physical memory alerts for windows servers

[dheer_thanu]

Hello Everyone,

In our organization we are getting frequent physical memory high usage alerts for Windows 2012/08/03 servers and this includes DB servers too. Our PM threshold level set to 80per and alert will be triggered following with Incident creation by monitoring team. Most of the services that consume high usage during the alert is JAVA, APACHE & TOMCAT. As we don't get proper update from application user's for our email which we inform them regarding the alert and kill some unwanted services. Some reply to increase the threshold but we are not ready to execute the change as recommended from Nimsoft agent.

What could be the memory allocation technique applies from VMware level and suggest me what changes need to be done to servers hitting 80per threshold. What could happen when the server hits 100per memory utilization ??

I can post more details if I'm not providing required details.

[bluefirestorm]

If I understand correctly, the high memory usage alerts is coming from the Windows 2012/08/03 VMs and not from the VMware ESXi hypervisor itself.

There are a couple of ways to handle this:

(a) increase the memory (i.e. virtual RAM) allocation to the VMs, for example, if current allocation is 8GB and you can increase to 12GB (if you ESXi hosts has enough free)

(b) increase the pagefile.sys sizes of the Windows server VMs

(a) might be harder to do especially if the VMs RAM allocation are already very high and hitting close to the ESXi host total RAM.

For (b), you can set it manually and set it up to 3x (depends on OS version) of the RAM allocation. The cost to this is disk space. If increasing the pagefile.sys size causes the free disk space threshold of the VM to be breached, you obviously have to increase the disk size of the virtual disks of the respective VM(s).

Have a look at this

https://support.microsoft.com/en-us/help/2860880/how-to-determine-the-appropriate-page-file-size-for...

Obviously you could do both (a) and (b). And if you do (b) first and (a) later on a VM, you might have to revisit (b) as the maximum size allowed would be different. You could also selectively do (a) on certain VMs or (b) on certain VMs, and both (a) and (b) on certain VMs.

Hope that helps.