Xen, KVM, Hyper-V, Oh My…And vSphere…Just Does Not Have The Same Ring As…Lions, Tigers and Bears?
January 28th, 2009
Virtualization Critical Comparison – Chapter 04
I have been spending a lot of time with XenServer 5, Hyper-V 2008, and of course VMware vSphere, over the last month or so. I expect I will continue to do this through March 2009. I planned to spend some serious time with KVM, but have not as yet had the time or resources to do so, yet. Fortunately or unfortunately, I have not been surprised by what I have seen and learned. My expectations have been met, things are improving in general, but maybe not as fast as I would prefer. No, don’t misunderstand me, I realize things take time. What I am saying is that I expected more from everyone, in comparison to everyone else, when new features would be added, not if. That everyone would steal more ideas from each other, and get these features started sooner than they have, in their respective platforms years ago, rather than in the coming year or so. For example, VMware now seriously is working towards imaging, or Microsoft now seriously working towards transparent migration. In both examples, VMware and Microsoft could have achieved these features years ago.
Here is my explanation of the key strategic issue per platform now, as a concept, not a specific feature comparison or analysis:
- At least the way I see Xen, given what I have researched and experienced so far. Xen seems to be dying. RedHat does not seem to know how to spell the world Xen, interesting since after all they were significant to the development and enhancement of Xen. I not sure why this is the case, Xen is a good platform, even if it does lack the features that make it a true enterprise solution, say compared to VMware. RedHat can not control the destiny of Xen, given Citrix being in the picture, so RedHat moving to KVM seems logical. However, Citrix I fear in light of Hyper-V, can not continue with Xen either.
- Microsoft, which should be called the Godzilla of virtualization, is just about ready to pop-up and trash everyone and everything that is not Microsoft. Of course, Hyper-V, yes, wonderful Hyper-V, well, it is a threat to everyone, but System Center Virtual Machine Manager (SCVMM) is a pain to install, a pain to use, and worse, it appears to me, to be very slow, slower than vCenter, which is significant. SCVMM is a pain to interpret, above all else, every single time SCVMM reports an issue it fails to explain it well, whatever the issue is. I love the…Host Needs Attention generic warning…most often this appears when an individual host needs a few hot-fixes applied, but now where in the SCVMM interface are the individual hot-fixes itemized? And of course SCVMM does not integrate into Microsoft Update, so you have no clue what the real specific hot-fixes issues are, if in fact the issue is missing hot-fixes. How many firms are going to purchase SCVMM or Microsoft Operations Manager (MOM) for Hyper-V? Not many. SCVMM and/or MOM are not inexpensive.
- KVM, as I said above, I have not spent enough time with KVM yet to have a detailed list of the good, the bad, and the ugly for KVM. However, as the late comer to this world of virtualization as we know it today, it has a very complex set of features and options that it must achieve to compete with VMware, and to some degree Hyper-V and Xen. The KVM open source heritage with support from IBM is a good thing, but can it really survive? Only time will tell.
- VMware of course is King Kong. Mature, stable, feature set deep and wide, most of the time there is obvious intelligence behind the design. But dang it, it costs too much. Given the current economic down turn, this is more significant than ever. I have read three (3) different proposals, in the last 10 days, from different enterprise architects from various places, that want to walk away from VMware, today, just because Hyper-V exists, and the single justification is, total cost, Never mind, that Hyper-V on clustering is horrible to setup and difficult to manage. That Hyper-V has some impossible single-point-of-failure-issues. Nevermind that supporting Hyper-V will be expensive given real and instance out-of-the-box limitations. VMware is in the clouds, discussing vSphere and its wonders. But are they watching the back door? ESXi free or not, is useless to anyone that has any significant scale of virtualization to implement and support. Any serious enterprise must have vCenter at a minimum. This makes Hyper-V look easy to implement at a strategic level, because it is just part of Windows 2008. Wrong. Hyper-V is an operational nightmare at a tactical level. Worse, if a firm is not going to purchase SCVMM or MOM, they of course will not be purchasing much if anything in or from vSphere. So where does that leave VMware?
At this point, I am sure most of you, are saying…What does any of this have to do with the title of this blog entry? I will elucidate. Well, the bears are gone; the bulls own the market right now, this economic situation, is innovation-to-support-expansion (ISE) death. True, I may have misled everyone, by referencing Bears, but…Lions, Tigers, and Bulls…Has no ring to it either. As for the Lions, they are the enterprise customers; they put up a good image, talk big, but behind the scenes, are doing little or nothing, everything is status quo, which is not good for any hypervisor vendor. Lions often have multi-year, enterprise licensing, so they will just continue as they have, using the product mix already licensed to continue in a tactical mindset. Thus the ISE projects are dropping off the RADAR like bugs flying too close to a zapper! The Tigers are the interesting aspect, they are the loners, singular, mavericks that will purchase what, when, and how they want, but over all, do not represent sufficient market share or scope to support the robust and extensive virtualization vendor industry.
I expect that the virtualization technology and related companies, especially the remaining 3rd party solution providers, for the most part will go the way of the Dot-Com firms collapse about 2000, we will see flood of failures, mergers, buy outs, etc., beyond the purchases that VMware has already made the last two (2) years. My view is that this chaos will last for about 9 or so months, with the avalanche hitting about June 2009, and thus going into 2010. This means that as Hyper-V starts to major in 2011, VMware will be in difficult situation. I would not be surprised, to see the greater hypervisor oriented virtualization market consolidate as well, to just two (2) firms, VMware and Microsoft. Leaving us, with King Kong, and Godzilla? Was there not a cult classic movie by that title in the 1960s? Yes, Kingu Kongu Tai Gojira. But I was always a Wizard of Oz fan, so I am sticking with Lions, Tigers, and Bulls, cough, Bears.
Entry Filed under: A Proper Virtual World
8 Comments Add your own
1. Armchair General | January 29th, 2009 at 3:57 am
What are your thoughts on the fact that VMWare is still after all these years a “one product company” now battling Microsoft who have come into the market with a product that is pretty much free?
History is littered with one product companies, with huge technical superiority to Microsoft’s (cheap or free) offerings, most of them are just a faint memory. Netscape versus IE v1,2,3 anyone?
I suspect VMWare’s life is now bounded at best (maybe even marginalized in future), where once it looked like a whole new world. MS will end up owning the market-share, and Citrix will do what they did with Terminal Services, build an enterprise business above the basic MS giveaway.
2. Jason Boche | January 29th, 2009 at 3:30 pm
If MS wants to give their products away, that’s their business. But I don’t understand the point of MS spending the development and marketing money other than for goodwill and philanthropy and maybe the propagation and perpetuation of the Windows OS. It seems like a lot of effort to convince people to use Microsoft Windows as a platform, when Windows already has a respectable market share.
No doubt VMware has a scary adversary. Personally today, I value VMware for their innovation and features that are not currently matched by any other virtualization vendor. Could that picture change in 3-5 years? Sure. That means VMware will need to stay a few years ahead of its competitors. While improvements can be made to the hypervisor, I believe it’s mostly cooked and for quite some time predictions have been coming true that the real selling points will be in the management software. It’s sink or swim for VMware, it’s competitors, and it’s partners.
One thing is for sure, virtualization is here to stay and anyone who can stay abreast with all of the virtualization technologies will be successful in the long run.
Jas
3. Schorschi | January 29th, 2009 at 7:29 pm
Jas, good points, thanks for the respone! I agree now is the key time for VMware, because the dragon is wake, if still a bit sleepy and uncoordinated.
Paravirtualization is gaining ground, and I think we are only a short time from seeing real significant abstraction at the chip level even beyond what paravirtualization is or will be. Call it BIOS level virtualization, processor level virtual hosting, where our entire model for a modern OS goes out the Window, or cough, Windows, ok, bad pun, but it was too easy to let it just cache.
4. Stephen Spector | April 14th, 2009 at 11:33 am
I find it odd that people think Xen is dying simply because Red Hat is moving toward KVM; although they continue to develop new solutions for Xen. The Xen.org community is very active and we are about to release our Xen 3.4 product with several new features developed by Oracle, Intel, AMD, Sun, Novell, Citrix, Red Hat, Fujitsu, etc. I encourage people who think Xen is dying to stop by the community and see that it is very much alive and continuing to grow.
5. Schorschi | April 15th, 2009 at 10:06 pm
The long term success of Xen has yet to be determined. Given that more is better, when it comes to competition, I would like Xen to be around for a while. However, when I polled my friends that are actively doing design and integration work for virtualization, and they run the gambit from small mom and pop to Fortune 10 firms. The exit from Xen as a real viable solution is factual, sad to say. Is my poll scientific, can I say more are going Xen versus than leaving, no, but nor do I feel that is the point. The point is, Xen captured a unique segment of the market that is the same as what KVM is focused on as well, the ‘I am less expensive than VMware’ scope and with the extensive muscle that RedHat has, Xen is under threat, never mind Microsoft Hyper-V playing in the VMware-costs-too-much arena. This cannot be ignored. Moreover, when I have friends saying they are dropping Xen in favor of KVM? That is real world. Is KVM with Virt-Manager as good as Xen? I would say right now, no, but as KVM and Virt-Manager mature together to a degree of integration faster than Xen did or has, because RedHat and IBM for example push its maturity to push on Microsoft and VMware? What does that say of where the future will be? For KVM? KVM has real strength in the Open Source community that Xen to some degree has lost. The buzz is key for open source growth, and the buzz is KVM focused now, Google makes this obvious as KVM gains critical mass. This combined with what Citrix has not done with Xen? I call that two strikes against Xen. Again, having worked with Xen, I am not thrilled with this situation, but I have to acknowledge it. Especially when I have clients and customers that ask me, what is the best solution for them, for the long term? I cannot in good faith recommend Xen for much longer, when the facts support Xen having a hard road and stormy path ahead as KVM dominates. Unlike Gartner, I don’t have a battery powered Crystal Ball, so if Xen does survive and trumps KVM, I will be quick to say so in this blog!
6. Vasquez | April 23rd, 2009 at 4:05 am
I have a hard time believing that people who claim that Xen is dying don’t have an agenda. The fact is that Xen is considerably more robust than KVM and Hyper-V. Additionally, while an ephemeral claim, I have managed Xen servers running a Citrix farm significantly faster than when it ran on ESX 3.5 (and in my lab, vSphere).
And now that Citrix has release all of Xen Server and its management tools for free, it would be irresponsible to not consider it.
Its true that VSphere’s feature set is far more robust than any of its rivals, you get what you pay for. But not everyone needs a Rolls Royce, and for those who don’t I would take Xen over Hyper-V or KVM every time.
7. George | May 14th, 2009 at 2:51 pm
I don’t agree about the complexity of clustered hyper-v server, have you tried clustered shared volume yet? It’s extremely easy! And reliable too.
I agree that the game and money is played on the management tools arena, even citrix follows the example.
8. Schorschi | May 14th, 2009 at 5:40 pm
At the time I wrote the original blog entry, CSV was not reality, however, as part of R2 beta, CSV is more than somewhere between vaporware and alpha, and maybe viable. As part of the R2 beta CSV can be evaluated, and I plan to do so. On paper CSV still feels like a cludge to me, not that CSV is not reasonable, more that the fail-over cluster model is weak compared to Xen or VMware. Microsoft uses what it has, but MSCS is showing both its age and its limitations in reference to Hyper-V, it is still a pain to setup compared other options. Of course Microsoft will improve MSCS as Hyper-V drives its maturity forward. MSCS in many ways has not changed since I worked with it and supported it in late/early in 1998/1999, on NT 4.0… Ouch, I just realized how long ago that really was!
Leave a Comment
Some HTML allowed:
<a href="" title=""> <abbr title=""> <acronym title=""> <b> <blockquote cite=""> <cite> <code> <del datetime=""> <em> <i> <q cite=""> <strike> <strong>
Trackback this post | Subscribe to the comments via RSS Feed