I have created 3 resources pool that I put everything at their default settings except the shares setting (1 RP set on high shares, 1 RP set on normal and 1 RP set on Low). From what I gather this setting is a relative number. Low is 1, normal is 2, and high is 4. But what I don't get is relative to what ?
They have no relatives. They are referring to available resource. So when your host becomes low on resource, VM's with a higher share get priority over other VM's. That's how it works basically.
Otherwise when your HOST is not low, ALL VM's get equal shares regardless of share value.
And what is the relation of the settings in the resource pool with the settings you put inside individual VM ? because if I go into each individual VM, I can also see the same settings. So what is the relation between the 2 ? does the settings in RP overwrites the ones in each individual VM ?
A pool is just that a pool of shared resource. So that pool is assigned an amount of shares to divide equally among ALL VM's in that pool. If you make another VM inside that pool more share, that VM may have a higher or lower priority but ONLY within that pool.
So, I have a total Pool of 47GHZ and 72MB & created 4 siblings RP. If go into the resource allocation tab I see that my two high shares RP get 36% and my normal shares RP get 18% and the low one get 9 %. Does this mean that my high RP get 36% of the total Pool when the times comes to fight for resources?
Close, it doesn't get a percentage of the resource, it gets priority OVER other VM's contending for the SAME resource. So the higher number wins, basically. Once those higher numbers get what they need, the next higher number gets their resource allocation, etc...
And since I put my limit to expandable and unlimited, there is a potential that one of my RP taking 100% of the shares if the others does not use it right ?
Limit and Unlimited is different than limit and unlimited of a VM. So if you have 3.0Ghz cores, for example. And you set your CPU to be unlimited in a VM, it doesn't mean the VM has access on an unlimited scale to ALL the resources in the Pool or host, so if you assign 1 CPU to a VM, and you have it set to unlimited, it's still capped at 1 CPU core speed, which in this example is 3.GHz. Just because you have 1 RP and 1 HOST and 1 VM, doesn't imply the VM can have access to ALL 8 cores, its still pre-defined at whatever CPU you assign to it.
You can further limit it's use of that CPU by setting this option to reduce the amount of CPU by setting an upper limit, so you can limit it to 1.5Ghz, which is essentially 50% of the CPU resource assigned.
Its a difficult concept to master, it just takes practice to understand it.
So even though my low one is at 9%, if that RP needs more resources and the other RP doesn't need it, he would "borrow" all 100% shares if needed right ?
It's not really 'borrowing' it's more priority, On a plane if you get on and the plane isn't full, you are free to change seats. If at the last minute the passenger with rights to that seat shows up, you have to go back to your seat or choose another free seat..... It's still his seat, until he leaves. Same with VM, if the other VM's aren't using the resource, and as long as the host has enough to give, ALL the VM's will get the seating.they want,it would be nice to move to first class if those seats were available.
RP are treated like VM's, they just represent other VM's. So a RP is a pool that the RP must divide among it's members based upon their priority. But the RP has to compete for other resources on the host or cluster with a higher priority, but I can see how you would classify this as 'borrowed' resource. It has to be returned if they want it back.