There is no licensing option (which isn’t extremely expensive) to allow us just to do simple things:

- Use vSphere vCentre (or whatever it’s called now) for centralised management.
- Do backups using standard off the shelf backup software (currently using customised ghettoVCB scripts).
- Possibly live migration (but not that important)

Vmware Essentials you say? No, that has a limit of 3 hosts. After that, we must pay £757 per CPU for the Standard Edition. That’s absolutely ridiculous for the minimal amount of extra functionality from the free edition.

What to do?

We use the servers mostly for development purposes, and much as I would like to, I don’t think I can get through £15,000 worth of software licenses… I had a hard enough time getting through a £5000 self-built SAN.

Essentials: GBP 380 (for 3 Servers), up to 6 CPUs
Standard: GBP 757.20 per CPU.

There is no bridge between the two – you can’t buy an essentials license and top it up with extra CPU licenses for more servers. You can’t buy two essentials licenses, only one.

This means standard costs at least £9084 for our current hardware. That’s just for the hosts though.

And then of course:

vCenter Server Foundation: GBP 1,140.00 – Ok fine.
vCenter Server Standard: GBP 3,795.00 – More than twice the price.

Plus, you must also pay for one year’s support on top of all these license fees. Not really feasible for a software business our size without a lot of planning – so we will have to struggle along with free edition.

The other major issue I have with VMware is – why the HELL do they keep changing the names of their products? Is this just to confuse people?

Created 6 vdisks in SCST and used these in rhev data center.
We tested the scenario twice using different multiple vdisk setup.

We failed miserably because rhev vdsm python scripts somehow
failed. (checked even the python byte code).

The problem is that even though rhev constructs VG using vdisks as PV
it still does check PVs consistency every time one creates new resources.
The consistency check seems to fail and the problem is within rhev.

Those Red Hat “engineers” are just dummies – they actually cannot do
anything other that asking questions like “have you contacted your SAN administrator?”

Naturally all the engineers which are NOT dummies are used for development
because they are not dummies (maybe)

The fix was to create one larger vdisk. And this was not funny.

//arl

DirectTalk === some cultures cannot handle directly to the point going semi offensive articles. proof reader told to me. but why should I care? [Sheldon Cooper]

Please help me to find some good RHEV deployment documentation.(not from Redhat)
I am strugling on the storage configuration part.

Regards,
Arumon

Also, does your KVM solution do clustering and transparent host migration? Hot-add CPU or RAM (which RHEV does _not_ do)? Host consolidation and power control (which RHEV “pretends” to do but actually doesn’t)?

If it were up to me, I’d consider Ganeti as well, but the features of VMware are QUITE nice..

But it will now be any day now and RHEV-M will be entirely JBOSS/Linux backend.

Those are the only advantages that RHEV has, besides pricing. Feature parity with vSphere 4 they’re promising for some time next year.

Plus, the RHEV-M runs on Windows, which for a Linux company is thunderously retarded. Hell, ganeti is more supportable from a Linux person’s perspective than RHEV-M, and if I could get ganeti working properly with gluster..

Really? Looks like Red Hat is kidding. HA depends on single point of failure – physical Windows box. Your whole infrastructure depends on it.