Hi guys,
Whilst I have been using the XEN hypervisor in our office/dev
environments for some time, I am preparing to deploy the latest XCP build for our
live/collocation installation.
To date I haven’t used any virtualisation at all in our
live environment as I didn’t see the purpose of it bar running a few
service/edge systems such as domain controllers etc. Our setup is relatively
straightforward with a three tier model:
·
Presentation tier (web servers, smtp, etc.)
·
Logic tier (business and operational servers)
·
Data tier (database and storage)
I am going to leave the Data tier non virtualised and virtualise
the logic and presentation tier. To date I have been able to simply add physical
servers to either tier to increase CPU/processing capacity or additional
redundancy. With consolidation of much more powerful hardware I should be able
to reduce the amount of physical hardware needed whilst increasing throughput
and reducing dead or idle CPU time.
So the example scenario is:
·
Both the Presentation and Logic tiers have 6
physical servers each (12 total) in a load balanced (shared nothing)
arrangement. These will be replaced with 3-4 physical servers running at least
two VM’s on each.
·
How do you set up the XEN XCP platform/environment
for dynamic workload structures?
·
If spikes come through via the presentation tier
how do you prioritise CPU/bandwidth to these servers/VM’s?
·
Likewise if the presentation tier is acquiescent
then how do you control it such that if the logic tier gets busy it can consume
idle CPU/bandwidth?
I have tried to find some relevant documentation or examples
relating to this type of setup but haven’t had much luck. If someone could
either help me out directly or point me in the right direction it would be much
appreciated.
Cheers
Chris
