1. Review memory requirements for your size environment. Any deployment using Control (NAC) or Analytics must use a medium or large install.Considerations for upgrading Extreme Management Center to version 8.x (formerly Netsight)
2. nsserver.cfg file is often not sized correctly.Extreme Management Center - memory settings -nsserver.cfg - for appliances running 8.x and higher
3. Use commands to look at top for CPU and memory utilizationHow to gather top CPU output over time to a text file
You can also look a 'free' command. If the used is >0
, then the swap space is using the hard drive as memory. This memory is not 'recovered' and so will always cause a potential performance condition.root@xmc:~$ free
total used free shared buff/cache available
Mem: 16423072 7381216 734416 828 8307440 8702836
Swap: 16777212 0 16777212
This also gives the physical memory total, the amount that is used and that is free. If the swap is used but less than free memory (Mem), then swap is not always being used. If it is greater than free Mem then it is being used often and will be a performance hit.
4. VM resources are constrained or over subscribed. Reserving resources may assist with this.
First, is overall platform resources oversubscribed?
Select the hardware host, and then look for the performance summary.
Here we can note the host is running over 80 percent for memory.
There is also Host->Monitor-Performance that may be more granular.Reservations:
This varies across ESXI platforms and management, but generally need to edit the platform when not running to get here.Hyper-V
has different settings Both for memory and Processors
memory settings should not be dynamic. Memory weight to high may give better performance
Here we can reserve a percentage of overall CPU to be 'kept' for XMC. We do not want to put a limit on it.