With windows vm's i can set a minimum ram and maximum ram and set the vm to dynamically allocate. When a windows vm gets under memory pressure it simply requests more ram from the hyper-v host and it's granted subject to physical ram availability and the allocation limits placed on the vm during setup. I have tested this functionality inside of UTM9 and UTM9 doesn't take advantage of this functionality. I don't know if this is a bug or by design. if by design then i'l turn this into a feature request.
Further testing mods i made with 8.9 just now: vpus = 2 25% minimum physical cpu allocation(1 core) 50% maximum cpu allocation(2 cores) 3 gigs ram 2 IPS instances
observed IPS behaviors: During snort startup the two processes pegged the cpu at 100%. the physical host cpu also jumped straight up to 2400mhz. This lead to a vastly quicker snort startup. Overall system startup improved as the host cpu spent more time at 2400 mhz than with previous cpu settings.
IPS performance: Full network performance now achieved. IPS now "pushes" host cpu causing increase in cpu clockspeed. While it is not to 2400 mhz it was up to 1750 mhz. Ram usage is now reduced due to not having to run 4 instances of snort.
Further testing mods i made with 8.9 just now: vpus = 2 25% minimum physical cpu allocation(1 core) 50% maximum cpu allocation(2 cores) 3 gigs ram 2 IPS instances
observed IPS behaviors: During snort startup the two processes pegged the cpu at 100%. the physical host cpu also jumped straight up to 2400mhz. This lead to a vastly quicker snort startup. Overall system startup improved as the host cpu spent more time at 2400 mhz than with previous cpu settings.
IPS performance: Full network performance now achieved. IPS now "pushes" host cpu causing increase in cpu clockspeed. While it is not to 2400 mhz it was up to 1750 mhz. Ram usage is now reduced due to not having to run 4 instances of snort.
