Archive for April, 2009

KVM, Now Things Are Moving, Well?

Virtualization Critical Comparison – Chapter 07

The last few entries to this blog have been focused on Kernel-Based Virtual Machine (KVM) technology with some metaphor parallels to Microsoft Hyper-V. Some may question why comments have been absent in reference to VMware? That is a good question. Well, to be honest, when grousing about KVM or Hyper-V, the actual best of breed gets a pass, for now. This will not continue, well not for long, once vSphere is out, there are bound to be a few things to discuss. Leave it to other blogs to comment on the latest buzz, the top new tricks, etc. This blog focuses on the things that from a strategic perspective are issues for virtualization.

A case in point and applicable to right now, KVM, or to be fair, the painful Virt-Manager. Having gotten a basic platform on KVM running, things are moving the right direction in the analysis of KVM. So this blog entry is going to be a bit shorter, KVM experimentation takes time, a lot of time. This has long been one of my disappointments about Linux based solutions; there is nothing quick about Linux the first time you walk down a new path to a new solution. Fighting the interface issues, the unexpected incompatibility issues, etc. Never mind digging through the tons of well intentioned documentation, but often dated material. Hate, absolutely despise, finding an article that solves one of my issues, only to get 2 pages into it, and realize it is only applicable to RedHat and not Ubuntu! To be balanced, commercial documentation is often dated as well, but at least there is someone that is paid to get it right, when it needs to be right, this not the case with open source of course. The ability to call someone and get things right, just is not possible with open source in the same way as RedHat or Microsoft.

Do not misunderstand; I enjoy digging into problematic scenarios. I am enjoying my ordeal with KVM. It reminds me when I first got an IBM PC XT, and wanted to run two monitors at the same time, months before anyone else, it reminds me when I got my Macintosh Plus, before any of my peers, and I figured out how to cross connect the XT and Mac, starting a small business as a University student, getting $22.50 an hour, when everyone else was getting $2.50, because moving data between the two systems was so new, I had the local market cornered. Dealing with KVM has that same feel. Unfortunately, I am not having the same enjoyable experience with Virt-Manager. I have been spoiled by VMware VirtualCenter, cough, vCenter. The issues with vCenter are real, the scaling and performance of vCenter plague its elegance and ease of use. Of course, vCenter regardless of its faults trumps Virt-Manager. So my expectations are out of whack, or unrealistic? Yes and no.

Yes, my expectations for Virt-Manager are unfair; Virt-Manager is wonderful, compared to what could be the situation, if Virt Manager did not exist. However, no, my expectations are not unreasonable, when considering that Virt-Manager should be closer to the competition, sooner than later. The interface needs some significant work to compete with SCVMM which is still not great, and vCenter. For example, the fact that you have to remove the network adapter completely from a virtual machine, to change the type of network, virtual versus bridged physical, is just not acceptable from an end-user perspective. Add and remove is not operational sanity when an edit option is obvious. To be sure, I really do want KVM and Virt-Manager to be successful. We need them to be successful. Once Microsoft and VMware do reach parity in functional feature set, only two outcomes are possible. Both are ugly for us, the humble end-users of virtualization.

  • VMware survives; they own the top 10 or 20 percent of the virtualization market, which are cost elastic. Any organization that needs and demands the best overall solution that is feature rich and continues to push the edge will continue to use VMware. Silicon Graphics in post production video, for example. Or even more significant, look at Apple Computer as an example. Apple re-invents its-self over and over; this is what VMware must do. In fact, I think Apple should purchase VMware and virtualize the iPod, iPhone, moreover consider the iTouch as a generic platform in this direction, so that many different hardware devices can function as iType devices. The first generic platform along this line, which has a common end-user selectable interface, which is the same on the HD TV, the Cell Phone, the DVR, even in the automobiles, which is not on a closed network, will, mark my words, crack the BlackBerry, and the hang it up for the iPhone.
  • VMware does not survive; Microsoft so dominates the virtualization landscape, that lowest common denominator virtualization platform cannot be surpassed once in place. This a proven model, why are there no new word processor applications? The barrier to the market is that critical mass cannot be achieved by anyone else other than Microsoft, simply because Microsoft is already there. This is why Sony Betamax though obviously superior, failed against VHS. However, there is one flaw in this situation for Microsoft, at least right now, for a short period of time, Linux based virtualization containers! Kernel based Virtual Containers (KVC) which does not exist as yet, but will once Sun Micro goes extinct, at least compared to what Sun has been in the past. Maybe RedHat will license Zones?

Why will KVC gain ground? Microsoft costs way too much per copy of operating system, and many fortune 500 firms are looking for ways to reduce operating system costing, licensing included. Microsoft forcing purchase of Datacenter is a panic response to how the slow of total operating system license purchases is hurting Microsoft. Microsoft is hiding the dark side of operating system downsizing under a new license strategy that will change again, and again. I will not be surprised to see Microsoft begin to lose market share to Linux in the server platform, faster than ever before. Do you not see this as well? Why is Microsoft pushing System Center so hard? Microsoft is worried. It is almost a joke, how desperate Microsoft sounds when trying to evangelize System Center, including Virtual Machine Manager, using Hyper-V as the vaguely successful crowbar to crack open the minds of CIOs to this refurbished concept. What exactly makes Microsoft think that if CIOs refuse to purchase vSphere they will instantly purchase System Center and Datacenter?

But I digress, or do I? One could say my disappointment is not KVM, but with Virt-Manager, why? Because I see Virt-Manager with its understandable but slow progress so far, as bad karma. We need the Linux based virtual containers, not Linux based virtual machines, and to achieve this, the management and ease-of-use aspect must exist sooner rather than later. I Hope the Linux community is thinking about this? Now is the time to be aggressive, when CIOs hate spending cash, cannot justify VMware or Microsoft big ticket expenses. I know for a fact, that every strategic thinker I know be they CIO, Strategic Architect, etc., just hate seeing Microsoft dominate the computing industry over and over, and real long term high expense that such a situation includes, no matter what Microsoft states today. Maybe some firms refusing to implement Vista should be a realistic indicator to the Linux community that the 1000 year era of Microsoft may be in trouble, about 30 years into the master plan? That a dynasty is posed to fall? Well, not quite yet. VMware has been unable to do it. VMware was in the best position to do so? VMware has had a great run, but has the same issue, cost. So is now is it time for Linux to live up to its highest potential at a reasonable cost? Time for Linux virtualization, KVCs, to leave the shadows, and take a place in the light?

7 comments April 14th, 2009


April 2009
« Mar   May »

Posts by Month

Posts by Category