is a buyout I can get behind. In my mind, Oracle buying Sun makes loads more sense than IBM
buying Sun (although I readily admit that many of you disagreed with my logic on that.)
The reason? There's not nearly the same overlap among products. Oracle will significantly expand what it offers now, rather than just offering more of the same stuff (e.g. servers) under a different name.
The big question, of course, is if this means Oracle wants to become a hardware company now (echoes of Cisco now wanting to become a server company, too). Oracle knows nothing about servers, of course, but it may have decided it's time to learn.
The question for the readers of this blog is about Oracle's virtualization plans going forward. Oracle has a good virtualization platform, although it's only for its own databases. (In fact, one of the ironies here is the joining of one of the world's most proprietary companies with one of the most open. Sun embraced open source years ago, while Oracle says "Hey, open source this, amigo!")
On the hardware side of virtualization, Oracle may have just bought itself a competitor to Cisco's new virtualization-tuned servers. Now Oracle/Sun servers can go out with a complete virtualization stack, right down to the virtualized Oracle database. Sounds enticing, no?
On the software side, Oracle gets Sun's virtual desktop infrastructure (VDI) offering, as well as a virtualization (and physical) management product, xVM Ops Center. Oracle didn't have anything substantial in the virtualization management space; now it does. One question to be answered in the coming months is whether Oracle wants to break beyond the narrow confines of managing its own stack, and enter the crowded virtualization management space into managing non-Oracle stuff. I'm also curious as to how much the Sun VDI solution factored into the sale; it's considered the next hot area of virtualization, so Oracle may have been trying to buy into the game. (It's also worth noting that Oracle gets VirtualBox, a great, free PC virtualization product).
What Oracle doesn't get, amazingly enough, is Sun's hypervisor, the still-unseen Sun xVM. One wonders whether work on xVM will continue, or if it will be integrated with Oracle VM. xVM, having been in development so long, has to be close to ready. Will it be released? If so, will it be re-branded by Oracle? And why would any company need two hypervisors that do the same thing (I'm differentiating here between different hypervisors, like Type 2 and bare-metal client hypervisors)? Questions worth asking.
Certainly, the IT public's main concerns will be the integration of the two companies on a high level. It will be just as fascinating, though, to see what in what virtual direction the new, much larger Oracle heads.
Posted by Keith Ward on 04/20/2009 at 12:48 PM0 comments
Stray thought about the huge news
today: One company that simply can't be happy is VMware. It's set to announce the details of vSphere
tomorrow, and was certainly hoping to have the news stage mostly to itself. Now it has to compete with one of the biggest IT stories of the year, a story which will dominate the week's coverage.
Sorry, VMware. At least take heart that the gigantic, multinational news staff at Virtualization Review is going to do its share on vSphere coverage.
Posted by Keith Ward on 04/20/2009 at 12:48 PM7 comments
I got an interesting tidbit of information from VMware today (not related to the big announcement
coming next Tuesday about vSphere.) I asked a VMware spokesman how much of the company's infrastructure is virtualized.
It's an interesting question, especially considering how Microsoft likes to tout that it eats its own dogfood. It is never shy, for instance, about mentioning that big Websites like TechNet are running on Hyper-V. The same question, naturally, is relevant to the virtualization leader.
The spokesman said that VMware's infrastructure is 97 percent virtualized, and "will be 100 percent virtualized within a few weeks." Most of that is on VI3, but a transition to vSphere is in progress right now, he said.
Considering that VMware is a major software vendor at the leading edge of the leading-edge IT space, that's impressive. Of course, it also helps that its VI subscription and support costs are zero.
Posted by Keith Ward on 04/16/2009 at 12:48 PM5 comments
Get ready for a big announcement from VMware next Tuesday. Although VMware hasn't made it official, everyone in the virtualization community knows VMware's going to be releasing the first broad details of VMware Infrastructure 4, which is going to be re-named "vSphere".
It's been a long time coming for most virtualization administrators, who have been using VI3 for several years now.
Make no mistake, though: this is going to be big. Very big. It will serve as the first glimpse of the specifics of the Virtual Datacenter Operating System (once known as VDC-OS), and provide a substantive roadmap for VMware's future in the clouds. Will VMware lap the competition again, or will the vision be misty and underwhelming? I don't know, but I do know there's a lot riding on this.
Some of the features expected to be included with vSphere (gotta admit, I do like the name better than VMware Infrastructure):
- Distributed vSwitches, including, presumably, the Cisco-manufactured Nexus 1000v vSwitch
- Fault Tolerance, which promises to improve on the already solid High Availability
- VMsafe, the third-party API for adding security onto vSphere. This feature has been the longest-coming of the anticipated upgrades
- Host profiles
- Thin-provisioned virtual disks
This is only a partial listing of what are expected to be the most significant features. But they alone are enough to get excited about.
On the downside, DABCC is reporting that there will be limited support for theVMware Management & Automation suite. That includes key products like Lifecycle Manager, Lab Manager, Site Recovery Manager and Stage Manager. Note that this hearsay evidence, and may or may not be true (although the authors claim to have heard it directly from VMware). If true, this would be a curious omission. I'd think that not supporting so much important technology out of the gate might suppress vSphere sales, in the same way that Windows Vista's initial lack of drivers and compatibility with older programs hurt its sales. We'll see.
Are you looking forward to vSphere? Let me know what you want most out of it. Also, if you have questions you'd like me to ask VMware, e-mail them to me or put them in a comment box below. I'm taking a briefing soon with the company on vSphere, and would be happy to pass along your inquiries.
Posted by Keith Ward on 04/15/2009 at 12:48 PM7 comments
Greg Shields, formerly a columnist with Redmond
magazine as well as Virtualization Review,
has one of the very best explanations I've ever seen on the architectural differences between Hyper-V and ESX in his blog
over at Realtime Publishers.
It's a balanced explanation, not taking sides, of the reasons Hyper-V is faster than ESX, as we found in our recent hypervisor comparison. In Greg's analysis, it mainly comes down to a question of drivers, and how they interact with the hypervisor layer.
So Greg, a challenge to you: now that you've done it for Hyper-V, how about a followup with XenServer and ESX? I know that Hyper-V are architecturally similar (although Hyper-V was not developed in conjunction with Citrix, despite popular mythology), but in our testing, XenServer beat both hypervisors in terms of raw performance. What makes it fastest?
In any event, Greg's article, like all his pieces, is worth a read.
Posted by Keith Ward on 04/15/2009 at 12:48 PM3 comments
The versatile Lee Pender, who writes a blog for Redmond Channel Partner
magazine as well as serving capably as Executive Editor for Redmond
magazine, my old stomping grounds, has an interesting blog item
up about a meeting he recently took with HP.
He quotes an HP virtualization executive who brings up a good point. From Lee's entry:
"In the physical IT environment, if you needed additional capacity for a marketing program you were going to run, you expected a month to get that capacity," said Bob Meyer, worldwide lead for HP virtualization solutions. "Now, with a server and hypervisor, I can set it up in minutes. I can set up a server much faster, but I'm really just pushing bottlenecks. I have a new environment, much more mobile, much harder to see."
He's right, too. The bottleneck is still there; it's just migrated, to use a favorite virtualization term. An analyst had this to say of the solution:
"We see virtualization and automation being inextricably linked. There's no way to separate the two. You cannot do virtualization without automation."
He's also right, for the most part. Small shops won't have many virtualization issues that can be solved with automation, but automation will be at the heart of enterprise virtualization management going forward. Without it, scalability bumps up against a very low ceiling. Anyway, the whole article is worth a read, as Lee's stuff always is.
Posted by Keith Ward on 04/14/2009 at 12:48 PM2 comments
Duncan Epping over at Yellow Bricks
has noted a security patch
that was just released for ESX 3.x. According to VMware, "A critical vulnerability in the virtual machine display function might allow a guest operating system to run code on the host." Hmm. Sounds slightly scary.
Download the patch ASAP if this is something that could affect your network. Of course, if you've configured Update Manager to do this automatically, you should be safe now.
Posted by Keith Ward on 04/13/2009 at 12:48 PM0 comments
Teradici Corp. just received
a round of "series C" funding that comes in at a tidy $17 million. It's a good sign for Teradici, and for the virtualization industry as a whole.
This isn't earth-shattering stuff, of course. Teradici is still a relatively minor player in the league, but I think they're poised for bigger things. The company makes PC-over-IP (PCoIP) hardware that dramatically speeds up remote desktop computing (Brian Madden did a recent video review of the product), which has obvious relevance to the virtual desktop infrastructure (VDI) explosion.
I would bet that Teradici's co-development deal with VMware had something to do with the amount of funding secured. For a relatively small company, $17 million is a lot of dough. "Series C" funding, for those unaware, is the final round of venture capital funding. It's for a company with shipping products that's looking to expand the reach of its business, including growing its sales staff and reaching out to overseas markets.
It's a safe bet that Teradici pointed to its collaboration with VMware to develop a software version of PCoIP as a primary example of its growth potential. Partnering with the biggest company in virtualization can go a long way toward opening the vault.
It also says something about the relative health of the virtualization industry that companies like Teradici can get significant cash infusions in the current economic conditions. VCs are confirming that market strength by putting their money where their mouths are.
Posted by Keith Ward on 04/13/2009 at 12:48 PM0 comments
I'm considering buying a laptop for virtualization and other uses. Of course, that means I need a real screamer. The virtualization I want to do goes far beyond the "load VirtualBox/Virtual PC/VMware Workstation and fool around"-type stuff. Yes indeed; I want to be able to build a virtual infrastructure and simulate a small virtual network, with multiple VMs and applications, interconnected to each other and the outside world the way it happens on a real network. All in software, of course.
Are you doing something like that for testing/certification/sales demo purposes? If so, I'd love to hear what your setup is -- what are the specs of your machine in terms of processor, RAM, storage, etc. This could also apply to a desktop, if you're using one for that.
If you have another way to try out virtualization products and use it in the real world -- remote simulation software, and so on -- let me know that as well.
Posted by Keith Ward on 04/09/2009 at 12:48 PM6 comments
So VMware is going to guarantee
that you get at least a 50 percent savings on server hardware, or their professional services are free. What to make of this?
First, it's a good thing for the industry. Sure, VMware isn't completely altruistic here. The company stands to make a bundle from selling Virtual Infrastructure 3.5 (now at Update 4 if you hadn't heard) and related upgrades, if the industry embraces this offer. But it also reduces some -- and maybe a lot -- of the risk involved for companies that want to virtualize, or do more with virtualization, but are hesitant to take on such a project in this economy. Knowing that they won't have to pay for the consultant work if they don't achieve 50 percent savings on server hardware should make it an easier sell to upper management, which is becoming harder and harder to convince to spend any money on IT these days.
Risk reduction is huge in a recession, especially one like this. That should lead to a greater adoption of virtualization, which can only be a good thing.
I don't imagine the savings part will be very hard to do. With most shops averaging somewhere in the neighborhood of six VMs per physical host, and the ability of most modern servers to handle double that number of VMs or more, chopping the number of servers in half should be cake. And the fact that VMware has limited the offer to shops of between 200-750 servers cuts down on the number of engagements.
Still, companies should look hard at the fine print of the offer. If you're CTO of a company that has done this, and you're convinced that your server hardware savings are only 40 percent, while VMware looks at a different set of metrics and says that it hit 50 percent, the fur could start to fly. Know exactly what you're getting into here. You also have to determine if the amount you could save on hardware will offset the cost of VMware's professional services, since if you reach the magical 50 percent mark, you will have to pay for the weeks or months of consulting that you're getting.
One question I posed in the news story is whether this program will change the perception that VMware is the most expensive virtualization solution (now, strictly in terms of upfront dollars, there's no doubt this is true; the question is whether the value of VMware's superior technology ultimately balances out that cost). I'm not sure it will, but at the very least it won't hurt VMware's reputation.
Would VMware be doing this in the absence of the recession? Director of Product Marketing John Gilmartin didn't have an answer for that when I asked, and it's an honest answer. Hypotheticals are hard. But it is a recognition of the current reality the industry is facing, and I'm glad to see VMware taking this step, even with the self-interest involved.
Will other companies follow its lead? We'll see; I'm convinced, however, that it will make many businesses consider VMware more strongly than ever. After all, they get a guarantee that the "other guys", at present, aren't giving.
What's your opinion? Would this make you consider VMware more seriously? Let me know.
Posted by Keith Ward on 04/06/2009 at 12:48 PM7 comments
the other day that our upcoming issue features a story about the display protocol wars -- you know, the battle to make stuff like RDP, ICA, SPICE, and so on work better, which greatly enhances the end-user experience that will be critical to acceptance of virtual desktop infrastructure (VDI).
A central thrust of the story is VMware's deal with Teradici to software-ize its PC-over-IP (PCoIP) offering. Currently, it exists as a hardware-only solution, but VMware wants to bring that kind of capability down to the software level. As you might guess, some folks are skeptical that any software version of a hardware device can get anywhere near hardware performance. Whether VMware can pull it off, or get its software PCoIP product anywhere close, is anyone's guess.
Since we just got the issue put to bed, I found it interesting that Brian Madden has a video up on his Website that takes a look at the Teradici device (the actual hardware, not the VMware software version). As you can see from the clip, PCoIP works quite well -- even video from the remote host looked good; not choppy or degraded, that I could tell.
It was an impressive performance from the Teradici device; if VMware can be in the neighborhood of this kind of performance, it could be a big win for its VDI efforts. A nice job by Brian, too.
Are you using Teradici? If so, I'd love to hear your feedback on what you think of it.
Posted by Keith Ward on 04/03/2009 at 12:48 PM1 comments
I had a very interesting conversation the other day with Greg Scherer, CTO for Neterion. Neterion just announced the release of a 10 Gigabit network adapter, the X3100 series
. Greg is a fascinating guy to talk to. Being the former CTO of Emulex, he's got some real insight into the virtualization arena.
One of the first questions I asked him was about the viability of the 10Gbps market. He estimates it at about 10 percent, and that seems about right to me. Fortunately for Neterion, they're not counting on this product to make or break the company. He said Neterion predicts steady, rather than spectacular, growth for the X3100, and the dual-channel version, X3120, coming out this summer. Neterion still makes its real money with standard hardware server and storage adapters.
It got me to thinking, though: how far along are your plans for 10Gbps? If you're looking at a highly virtualized environment, it's probably something you've either instituted or are planning for right now. If you're going to be getting into other virtualization technologies like virtual desktop infrastructure or disaster recovery, you're likely looking hard at it, too.
So let me know what's happening with your infrastructure plans, either below or via e-mail. Tell me what bandwidth you currently have, what you expect to be virtualizing in the near future, and how you'll be upgrading your bandwidth to accomodate the changes.
Posted by Keith Ward on 04/02/2009 at 12:48 PM3 comments