Proper Network and Infrastrucutre Sizing for Large VM's

Proper Network and Infrastrucutre Sizing for Large VM's

Hi all,

I have a discussion / question for you,

We currently have around 100 hosts and 2000 VM's across North America.  Our current virtual infrastructure is designed for standard server builds.  We have been given new business requirements to be able to build out and sustain a very large / resource intensive VM solution.

Each VM will be running RHEL 6.1 , they will have allocated 96 GB of Memory and will be run on a DL980 with 2 TB of Memory,  each DL980 will be running approximately 20 of these large VM's.

My question is around proper design for such large VM's ... For example here are a few questions :

What size / speed would my management network need to be?

What size / speed would my Vmotion Network need to be to migrate such large memory consuming vm's?

What size  would my host scratch volume need to be ?  does this even change ?

Do the RHEL VM NIC drivers allow the VM's to communicate at 10 GB's ?

Other foreseen issues with hosting such large VM's on this type of infrastructure ?

If you have any good input or experience hosting very large and memory intensive VM's on similar platforms,  please let me know how you addressed the above questions and sizing requirements.



Version history
Revision #:
1 of 1
Last update:
‎02-08-2012 02:12 PM
Updated by: