Virtualization Automation – A Proper Virtual World

The Rise and Fall of KVM?

March 19th, 2009

Virtualization Critical Comparison - Chapter 05

As I listen to the Halo original sound track, a sense of shock and ah creeps up on me, in tempo, to the music. What Halo sound track? Hey, I am a geek, get over it! But moving on to the topic at hand, I promised that I would deep dive into KVM, as I have with Hyper-V, and parallels between these two virtualization solutions, at least so far in my analysis are similar as well as not on some key points. I have by no means finished my dive into the virtualized ice supported by Hyper-V and KVM, but here is what I have in my notes so far. Expect this to change over time, as I get deeper into the ice.

• Hyper-V is not ready for the enterprise, large scale enterprise. This is not to say it cannot be made to work, but Hyper-V today is not a clear winner compared to Xen, vSphere, or even KVM.
• KVM is still difficult to install and configure when one takes into account that its implementation is not as easy as KDE across multiple distributions; I found RHEL and Ubuntu to be easier than Gentoo for example, even though Gentoo is my favorite Linux distribution. This is not to say that any of them are very hard or just easy, only that the multiple distributions make KVM standardization a bit hard, and digging through all the Google hits a pain in the posterior.
• SCVMM in my view has great potential but fails short of the mark. It is slow. It still has issues with complex Active Directory designs, and disjointed domains. It does not eliminate vCenter, only links to it. Notice I did not use the word, integrate, SCVMM does not integrate vSphere, because it just glues its-self to vSphere and makes vSphere management, with its limited scaling, look like SCVMM, yuck. I really do hate the SCVMM interface. Microsoft has bundle System Center in such a way as to target vSphere management suite, this is interesting, but very expensive no matter what Microsoft claims. Management of virtualization is where everyone thinks the money to be made is at. Idiots. Chasing dollars and not improving the total solution is just as bad as adding new features in a half-ass manner.
• KVM management, this I must admit has been struggle for me. I am not used to using a GUI in Linux, and the fact that I really have to use a GUI to be effective in management of KVM, is fine, just a bit of culture shock for me. Odd, I accept a GUI in MacOS with no worries, I work with Microsoft Windows, which I still feel is a rip-off of the original Machintosh OS, but balk at a GUI in Linux? Never liked Gnome, that may be coloring my perception.
• KVM is still immature with its support of Windows, I am attempting to run Windows 2008 R2 Beta, and having no end of issues, I suspect many of my issues are learning curve issues with KVM management tools, virt-manager, and the GUI issue as noted above. I really need to setup a Ubuntu desktop, or server with KDE, and that is the plan. Even when I reverted back to Windows 2008 proper, the issues remained, so it is a KVM issue from my perspective at the moment. It is interesting that every example of Windows running in KVM is desktop class software, Windows XP to the greatest extent, and a odd instance of Vista here and there. What does this tell you?
• Hyper-V no matter what Microsoft says is still not on par with vSphere when it comes to a clustered file system. No this is not an argument about NFS versus VMFS, even with the SCSI reservation scaling issues and such on VMFS, VMFS kicks everything else to the curb right now. Sure NFS is gaining, and may at some point own the shared IO storage space, but not right now. Clustered Shared Volumes (CSV) feature in R2 is going to be a disappointment to many. It is nothing more than NTFS with a few filter/wrapper drivers, so it is not going to win any awards.
• Some time ago, I told Microsoft to walk away from NTFS shared, and just adopt NAS, NFS and iSCSI and call it done, but Microsoft has problems listening to the customer at times. Microsoft needs to get to NAS with Hyper-V quick. Much quicker than VMware did. NFS and iSCSI, thought I am not endorsing either here, significant threats to classic fiber-channel, never mind FCoE, which is a solution looking for problem until it is finalized and officially adopted by at least a few big vendors. NetApp is yelling from the mountain top various points of how NFS scales beyond VMFS, duhe. But until everyone has 10GB dedicated storage networks, FC will always have its place, but it is threatened, no doubt.
• Hyper-V with its very immature network model, that is not even close to vSphere or Xen, is another weak point, that is going to plague Microsoft for months if not years. Microsoft needs to implement a MPIO like model for Hyper-V virtual networking. Why Microsoft did not do this in Microsoft Virtual Server, is a mystery as well, this was an obvious and clear winner for Microsoft.
• Why am I referencing the above Hyper-V issues, because KVM addresses these out of the box, in fact, the further down the rabbit hole I get, the more I see KVM not as threat to vSphere but to Hyper-V. Interesting no?
• KVM has one other significant win, it is lean and mean, it is a virtualization container model, similar in strategic approach to Solaris Zones or LDOMs, and as you all know, from my past discourse on the topic of virtualization containers, I see this as the first step to operating system instance reduction. We need to reduce the total copies of the operating system running in parallel in virtualization, now is the time, everyone!

I am slowly coming to the realization that vSphere may be irrelevant, based on cost, and Hyper-V and KVM will drive right at each other, Xen is always road kill, not because it is wrong or weak, but because it has not momentum behind it. KVM has something that VMware has lost, and Hyper-V is trying to gain…Energy! No, Passion. When you read about what developers are doing in KVM, and how they discuss issues, there is emotional if not a spiritual aspect to their oration. Excitement and motivation in their words, and this is critical to the future success of KVM. I perceive the significance of this having once been a down and dirty developer in my college days. It is not the high caffeine loading, or the potential profits, that drive developers, it is the bragging rights around the water cooler, cough, the instance messaging groups or dig-it or even tweeter, about the latest kewl trick that is rad if not sick, that was added to KVM, which has everyone jumping online to download KVM. The Linux architecture fans are just itching to nail Microsoft, and as a freebie kick VMware at the same time. I will continue with my deep dive of KVM and comparison to Hyper-V, and I expect my perceptions to change or surprise maybe re-affirmed? But I am distracted.

Changing Gears for a Moment…

About 40 or so people, some very close friends that are now unemployed, all in the IT industry, that I have worked with for years if not decades and knew will, in a work association context, have been let go, laid off, etc. This weights upon my thoughts often now days. The aspect of this current economic situation that saddens me is this, the pain I see in others, in the looks on faces, the words left unsaid, living through the economic instability cut lose in the wind if you will, through absolutely no fault of their own. It is difficult for me to endure. Life is not fair, to be sure, not even life in the IT industry. Nor do I blame those that make these ugly decisions; it is a thankless situation for all. Decisions have to be made.

But still, some of these people, friends or otherwise, I know for a fact, where ranked very high in their respective organizations, provided excellent service to their respective employers for decades, where often recognized and rewarded, and still the axe came with merciless speed and disregard of achievement, skill or even talent. It begs the question…what criteria is used for deciding who is left standing, and who falls to the axe? Moreover, I ask…how many times does the axe fall, before any given firm loses competitive advantage? How many times does axe fall, before any given firm loses talent and thus knowledge that is gone forever?

What is the true definition of profitability and strength of a firm? What drives strategic success? Tactical results of course. It is skilled and talented people, in-house, that do real work. Retaining those that make your respective firm profitable is strategic and tactical, the people that work where the rubber meets the road, not management, not directors of the board, of course not the investors, and dear God, not even CEOs. How many talented people can a firm retain, if it let the top 10%, no, top 5% of its management structure go?

Furthermore, is it not ironic, that we call it in-house resources and out-sourced resources? Of course, the most straight forward descriptor to the opposite term for in-house, the true antigram to the term in-housed resources, is well, honestly, is out-housed resources? But I guess out-housing does not have that positive executive level ring or spin to it, now does it? But it does have a certain a semantic parallel, right?

Entry Filed under: A Proper Virtual World