Resource Pools Question

[psynophile]

I'm fairly new to the VMware Infrastructure. I've read a lot of the available documentation, but I'm having a hard time getting my head around Resource Pools. I was wondering if someone could point me toward some documentation which included examples of how and why Resource Pools are used in a real-world production environment. In the meantime, if someone could answer the following questions, I'd appreciate it:

It seems to me that RPs were created to delegate resources according to how many resources an organizational unit needs. For example if a unit needs 5 machines running at 2mhz each with 2 gig of ram each, then an appropriate unexapandable RP for that unit would include 10mhz and 10gig of ram and if any machine would be created that would exhaust those resources, the machine would not be able to be started. Is that correct? If so, then how can the RP manage how many CPU mhz each machine is using? If all 5 machines were started at once, wouldn't the 10mhz limit be quickly exhausted? Would the machine be halted/shutdown? Or would the machine just run with >less< mhz?

Do RPs affect DRS or HA? How can it? Since DRS moves a machine if it's using a set amount of the resources on an ESX machine, how do RPs come into play if they are universal to the cluster?

Thank you very much

[marcelo_soares · Accepted solution]

psynophile,

Pertinent questions. RPs are used at the "eral world", yes, but I've not seen too much real implementations using it. But all that I have seen, they work very well for limitation and reservation purposes. Answering your questions:

"any machine would be created that would exhaust those resources, the machine would not be able to be started.": this depends. The limitation will be strong for the RAM (the sixth VM will not power on), but for CPU the ESX will try to share the 10GHz (I suppose the typo at the MHz) with all the VMs on the RP - always counting that you can provide each machine with a CPU reservation, so if the reservation exceeds the limit, then you will not be able also to power on the Nth VM.

If so, then how can the RP manage how many CPU mhz each machine is using? If all 5 machines were started at once, wouldn't the 10mhz limit be quickly exhausted?: This depends on the CPU usage of each VM. The performace will be degraded IF the 10GHz limit is not enough for them.if you have 5 database servers inside this RP, they probably will present some performance degradation. But will never halt/shutdown.

Do RPs affect DRS or HA?: As the RPs are on a cluster level, they don't. DRS will move VMs from one ESX to another depending of the host CPU/RAM usage. HA will be not affected, only that your reservations can reduce the amount of failover servers you configured for it.

More docs on it: http://www.vmware.com/pdf/vi3_35/esx_3/r35u2/vi3_35_25_u2_resource_mgmt.pdf

Hope this helps,

Marcelo Soares

VMWare Certified Professional 310

Technical Support Engineer

Linux Server Senior Administrator

[psynophile]

Thank you for the comprehensive and informative answers. They are exactly what I was looking for to complete my understanding.

I've read through those documents, but real world experience tops and I had a difficult time visualizing how to plan my new set up. Yes I meant ghz, sorry about that.