Shawn Roberts – Vancouver. British Columbia

Market share is a hard fact to beat, and VMware has that hands down. I am a Microsoft junkie and I hate their Hypervisor.

I have few Linux skills but I can setup huge farms of ESXI easily and I can cover all the bases from SQL to AD along with hardening… Can’t say I have no skillz… Just suck at Linux is all…

After all, what would VMware be without POSH!

Many people detest RHEV (2.2) because IT STILL COSTS MONEY, it involves a WINDOZE platform as the “manager” (which can change to a Jboss App with v3.0) and simply because it exists (bigotry, indiference to the new kid on the block)! Ubuntu really should have spread the word early on being the first adopters of KVM but that is a different story…

Unbeknownst to most — even to Linux Pros — is that their RHEL A/P License already entitles them to the ingredients of building a highly customisable, open, single vendor stack (no blame game) HA Virtualization that can rival all those Point and Click “VCPros” ;^)) RHEV *just* makes the management and orchestration of KVM better as easy as the 3rd grader that manages vSphere ecosystems. There are others out there that makes KVM beautiful i.e. Proxmox or Enomaly but that is a whole different discussion.

And with CentOS around, does vMware have something TOTALLY free that can provide all the HA features of KVM + libvirt? None! I too have been a long and loyal vMware Server and ESXi fanboi until I got wind of KVM and Libvirt being free and available on CentOS and found out I can build a highly available Virtualization Cluster… it was a no brainer, I converted all my vmdk based VMs to KVM, ditched the pesky vmware tools and my clients are just as uber happy with spending NIL to have “enterprise class” virtualization…

Again, vMware empires and fiefdoms in Corporate IT may very well be well entrenched (weary of their own skin) – it may be a tough challenge even to make mention of KVM./RHEV. In my workplace – our vMware Chief of Fiefdom even was uber worried and always worry by the mere mention of RHEV/KVM…

Live and let live — the world is not that dumb ya know!

There is place for everybody, even in enterprise virtualization, despite what this blog try to preach.

You could in theory have a perfectly servicable clustered virtualization solution in only opensource software, I have in fact argued for same using OpenNebula. “corporate IT” management said no. Guess what? The “corporate IT” market is stupid and inefficient, but it’s also far more lucrative than the opensource market.

For the market it’s trying to sell into, RHEV just fails. If you can do ON/Ganeti/etc, you wouldn’t need RHEV in the first place, so what’s the point?

This is absolutely TRUE. One does not need RHEV to build a KVM High Availability Virtualization Cluster. One simply has to use RHEL 5.4/5.5 or 6.0 with the Advanced Platform subscription which provides all the “ingredients” – RHCS, GFS, CLVM, KVM and libvirt. I personally do not like RHEV as well due to the way it names the data stores — BUT with a customised brew – one can use GFS2 and CLVM datastores – firnedly named and aptly customised for performance, etc.

And if you use Redhat’s downstream CenTOS release – ZERO COST! Now – that can’t be beat. The same holds true for Ubuntu.

With RHCS (Redhat Cluster Suite) as the HA backend to KVM, one can even easily configure the Virtual Machines as a clustered service for easy failover, consolidation (power savings) and load balancing accross physical servers. And the beauty about it is the Physical Servers does not have to be the same make/model/CPU type as I can fail over between AMD and INTEL cpus! If one does not wish to make Virtual Machines as RHCS services — it is quite easy to use say NAGIOS or a hombre script to effect load balancing and automatic failover on proactive impending Physical Server Failures. And since CLI via virtsh is there – utmost flexibility in management and customisability..

My site also needs Virtual Linux servers that need more than 8 vCPUs than vSPhere VMs can provide. My KVM Guests can also be configured to have direct access to a FC HBA or NIC that has less drag and better performing than the other hypervisors. ANd did I mention – I do not need to have a vMware Tools that needs upkeep and a recompile whenever I upgrade my Linux virtual guest?

So c’mon vSphere/vMware bigots — I am not even pitting RHEV which can just be better, I am simply pitting built in features against vSPhere… And come to think of it – here’s a newcomer – and already making the vSPhere folks and their corporate fiefdoms tremble…

One oft asked question — why are there no Big (North America) Names yet using KVM/RHEV? I think it is because of well entrenched vSPhere fiefdoms who are NOT Linux/UNIX pros and who are protecting their turfs and not necessarily corporate interests. But its coming..