Virtual brownfields: Microsoft's push to woo the VMware masses The virtualisation war is heating up. With Server 2012 Microsoft is at last bringing a viable platform to the table and Server 2012 R2 looks set to eclipse that. You will get different numbers depending on who you talk to but the general consensus is that Microsoft has managed to grab about one quarter of all new installs. It …
I currently run Hyper-V and VMware in my home lab, I've been running VMware at various places of work for several years. They're both on a rough par with each other now, some better features in each when compared to the other. VMware has USB support and ability to dynamically change hardware (disks, in particular) with the machines on. Hyper-V has a pretty good memory management system which seems to work much better than VMware, IMO.
Really the choice is down to money, how much does it cost your Enterprise? The really, really good, killer feature of Hyper-V is that now it's a credible Enterprise player, there is now a big stick with the word "Microsoft" written on it that you can use to beat EMC with. Likewise there is a big stick with "EMC" written on it that you can beat MS with. Competition means there are customer wins.
If you use Hyper-V Server 2012 (as opposed to Windows Server 2012) then the OS is completely free. Trevor doesn't seem to be aware of the difference. In an enterprise you would normally user Hyper-V Server due to the greater efficiency / reduced attack surface / lower cost compared to Windows Server.
In terms of technical advantages, Hyper-V has better scalability / IOPS performance than KVM, and is on a par with with VMware ESXi 5 - but at a much lower cost. Things that are very expensive on VMware like replication are included free in the base (free) Hyper-V server product. Hyper-V + management tools has a much lower TCO.
The main technical advantage of VMware is memory deduplication, but RAM is cheap, so it's not that big a deal.
Worth noting also that VMware ESXi 5 has had well over 200 vulnerabilities - Hyper-V server 2012 has had less than 10....
And if you look at the Linux console based ESX, then of course it's much higher: http://secunia.com/advisories/product/25985/ - over 700 vulnerabilities!
And if you look at the Linux console based ESX, then of course it's much higher: http://secunia.com/advisories/product/25985/ - over 700 vulnerabilities!
And if you look at the Windows underneath, then of course it's much higher: [integer overflow]
If you're going to pull out stats, stick with the one (direct competitor) product...
The direct competitor of ESXi is Hyper-V Server 2012. Windows Server 2012 doesn't add anything to the Hypervisor layer.
However, If you want to compare Windows Server 2012 OS, then the whole of this still has fewer vulnerabilities and fewer unpatched vulnerabilities than both ESX and ESXi.....And of course an enterprise Linux distribution with KVM gets into an very large number of security holes....
http://www.secunia.com/advisories/product/42761/
If you're going to try to debunk figures about vulnerabilities, it's probably best to actually cite a source rather than resort to sarcastic "Windowz it teh loozer" type sarcasm. I'd be genuinely interested to know if there are a significant amount of vulnerabilities in the Windows components which are required to run Hyper-V, I personally suspect there aren't, but if someone can show there are I'd like to know.
Given the cost of SCVMM (it's NOT free), there's still an important alternative for smaller shops (up to 3 ESXi Servers, 6 sockets) which includes vCenter Server - Essentials (Plus). Note that you get lots of functionality (Replication, Backup, HA, vMotion) already with the Essentials Plus version. So VMware isn't necessarily much more expensive then Microsoft.
In general however, licensing is just one of the costs of running a platform. Just like a number of other people noted; management is something totally different. And there things might turn out to be radically different.
Also, migrating to a new version seems to be a real pain with Hyper-V, whereas it's relatively easy with VMware.
"Given the cost of SCVMM (it's NOT free)"
SCVMM is just a GUI though - you can do everything that it enables via PowerShell for free....
Essentials Plus doesn't give you Storage Vmotion, which is a big omission. It also costs almost £4,000 - far more than the Microsoft System Centre licences for 3 hosts. So even this is a LOT more expensive than Microsoft.
Migrating from previous versions or from VMware is really easy if you know what you are doing. See http://blogs.technet.com/b/keithmayer/archive/2012/09/13/migrating-to-microsoft-hyper-v-server-2012-from-2008-r2-and-vmware-winserv-hyperv-itpro-vmware.aspx#.UdrRLKPTWUk
Call me cynical, but perhaps there's a cartel going between the major hypervisors.
Let's face it, the cost of the licences (which may appear to be zero) is an awful lot less than you'll spend on consultancy and projects to do a full scale migration.
So while ultimately it may look like you'll "save X over Y years" by doing the project, you've spent 90% of the costs in year 1 with the vendor instead of spreading it over the full Y years. Factor in the implications of Moore's Law and trying to figure anything "hardwarey" more than 18 months is just guesswork. But that won't stop the bean counters.
"""With Server 2012 Microsoft is at last bringing a viable platform to the table and Server 2012 R2 looks set to eclipse that"""
I love that, this is how always MS does things, the current version is the best ever.
Hyper-V in Win 2008 and 2008R2 is a turd, all I see in 2012/R2 is more shiny tools, and no convincing reason to move.
Competition is good. We would like to see more competition based on making the best product and far less competition based on who can do the best job locking customers into their little walled gardens.
Personally I wouldn't trust System Center to watch my dog, much less a heterogenous data center.
But if we're really heading towards a world where the hypervisor itself doesn't matter, then the smart money would be on open source virtualization. It's the typical scenario: the big names fight over who has the most features, while the open source offerings do the basics better than anyone else. It's difficult to imagine a hypervisor more efficient than KVM, and the management frameworks (such as ProxMox and oVirt) are starting to match most of the feature sets "enterprise" customers are looking for.
Thanks, a good little intro into the possible pleasures or pain of switching virtualization hosts and migrating VMs. One thing I would like to point out is that, in the enterprise arena, we often ask for a whole lot of features we simply never use. Case in point, we are currently a VMware shop and comparing it with Hyper-V and KVM, but we are making some silly comparisons, such as live migration of VMs from one visor to the other. In practice, when we make any changes to a system image, we run it for at least three months in T&D before we take the new version live, and whilst our virtualization gurus are getting all uppity about live migration the reality is we simply would not do it live. We're much more likely to run a clone on the new visor in T&D and then just switch the prod feed to the clone, keeping the old production image available in the background in case we need to roll back. Later, the old prod system might be recycled to be the new T&D environment.
As was noted in the article, few could justify why one "would" move from VMware on GNU/Linux or other Unix-like operating systems to Microsoft's hyper-v on Windows.
it is my clear understanding from several reports via IBM, Oracle and other large IT firms that the fly in the ointment for Microsoft in convincing - with any facts - clients to transition to their virtualization are the many disadvantages of Windows Server when compared to RedHat Enterprise Linux and other well known Linux and UNIX-like OS which have proven unequivocally up until very recently to be verified more reliable, scalable, flexible and critically more secure than even Microsoft's latest offerings.
Several comprehensive evaluations in the US and Europe, in financial and scienific research sectors show that more established Linux solutions remains substantially more viable and robust against any Windows OS, and that transitions to Microsoft base would carry serious and great risks.
Furthermore the Microsoft solutions are significantly more costly and complicated for shops not "all-Microsoft" than VMWare and KVM virtualization under any circumstances.
God help the poor slobs who converts as result of Gee Whiz marketing efforts of Microsoft and it's empire of eager partners , only to discover there-after that they came up short in scalability, flexibility, performance, security and lowered value (greater outlay) of such a move.
"only to discover there-after that they came up short in scalability, flexibility, performance, security and lowered value (greater outlay) "
I'm the first to tear Microsoft a new one on pricing and licensing shenanigans, but "scalability, flexibility, performance and security" are all areas that you are going to have a hard time convincing me that Microsoft is not at par with - or is - the industry leader. This isn't the XP era. Microsoft are a legitimate player.
This isn't the XP era. Microsoft are a legitimate player.
They friggin' should be. It's not like they didn't take their time getting there. I can see the OP's point, though - one of the earliest signs of Windows problems in the world of reliable servers was just how much it took in the way of resources to approach the reliability of a simple, boring standard Intel box with a free Linux distro (I mean, a full CLUSTER to go anywhere near the sort of uptimes for a stupid Linux box supplying the same print & file services?).
I remember the other IT guys gasping when they found out why we delayed upgrading a box in our setup for a few days: we decided it would be cool to give that box the chance to reach a YEAR uptime (and that wasn't even on proper server hardware :) ).
That's actually what brought Linux into the enterprise in the first place: sysadmins slipping in SAMBA boxes to get some reliable service where Microsoft just couldn't deliver.
If they now can, fine, I'm just not vastly impressed with how long they took to get there.
Do you actually believe all that? I've worked in financial services IT for about 15 years and now work for a major software/hardware vendor and it appears to be to be utter drivel.
NT4 was good enough for most people, we got about 3 months uptime before elective reboots on our servers at the FTSE100 FS company I worked for at the time.
Anyway, uptime is a red herring, it's prevention of un-scheduled downtime that you want to watch out for. I am aware of companies who rebooted their NT4 boxes every Sunday, after the backups had finished and consequently had next to zero unscheduled downtime.
"....uptime is a red herring, it's prevention of un-scheduled downtime that you want to watch out for.....rebooted their NT4 boxes every Sunday, after the backups had finished and consequently had next to zero unscheduled downtime" Unfortunately, it wasn't just NT4. Prior to Y2K I worked on a contract where we had a similar "workaround" for some pretty bad C code on UNIX. After about ten days of running on Solaris or hp-ux, the application would crash the server. To meet the production schedule, we put a cron job in for reboots every Sunsay at 3am at the end of the week's full backup. Of course, once it was in production, there was a resistance from the company to invest in going back to investigate and re-coding to fix the issue, and the C coders were all contratcors and not willing to put the time in for gratis, so in the end the problem was never fixed. I always remember it with a smile whenever some UNIX fanboi gets all misty-eyed over "uptimes".
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
