Re: clustering solution
To second Ed's mention, VMWare Infrastructure is really going to match
all of your needs. It will take you about $3000 to get started with
it, but if you put that cost alongside the hardware you'll be using,
it starts to make sense.
ESX and ESXi are installed "bare-metal" on each physical server.
These are based on a Linux kernel. ESX is a full-blown version for
about $500 and ESXi is free, but supports less hardware.
Virtual Center is what controls the ESX/ESXi servers. For some reason
this only runs on Windows Server, which is odd considering their
commitment to Linux elsewhere.
So you setup your Virtual Center and install ESX/ESXi on your physical
machines. The thing here is that you can just "throw" hardware at it
and build your virtual machines. Virtual Center will look at what
resources is available and pick the best physical machines to move
your virtuals to, automatically (without downtime). As you add
machines to the cluster, VMWare will move virtual machines as needed.
Another great feature is that it will try and consolidate your virtual
machines on as few physical machines as possible. The unused physical
machines? Placed into hybernate/sleep mode. That way, instead of 20
machines doing very little, you end up with 12 machines working hard.
As more physical resources are needed, Virtual Center will wake up the
sleeping machines as needed and migrate with zero downtime.
Of course, there are overrides to lock virtuals in place or keep
virtuals from being moved to a physical host (for example, if you're
only testing new hardware and don't want it to be swamped with mission
critical virtual machines).
-John
On Fri, Sep 19, 2008 at 2:04 PM, Stephen Kratzer <kratzers@...> wrote:
> All,
>
> I'm looking for an enterprise-class clustering and virtual server solution
> that would allow me to:
>
> - aggregate the resources of a large number of physical servers into
> allocatable pools of resources including CPU, memory, disk space, etc.
> - create a large number of virtual servers which can each be allocated, with
> fine granularity, a fixed or variable amount of resources from each pool
> - replicate virtual servers across clusters so that any cluster, in addition
> to running its own virtual servers, can act as a secondary, backup server for
> another cluster (with failover happening automatically in the event of an
> entire cluster failure or even a virtual server failure)
> - easily manage clusters and virtual servers
>
> Does anything currently available even come close, or is this wishful
> thinking?
>
> Stephen Kratzer
>
>
>
>
> _______________________________________________
> CPLUG mailing list
> To unsubscribe send a mail to cplug+unsubscribe@...
> Archives: http://dir.gmane.org/gmane.org.user-groups.linux.cplug.general
>
>
--
--
John Hogenmiller - ytjohn@...
Used for mailing lists - sporadic response
RSS Feed