Once an IT shop embraces virtualization and provisioning takes off, virtual server sprawl can quickly require third-party virtualization management tools. What types of virtualization management tools are available to enterprise CIOs, and how affordable are they? In this podcast, Karen Guglielmo, executive editor for SearchCIO.com, interviews Forrester Research Inc. analyst Galen Schreck about the latest virtualization tools from both traditional systems management vendors and startups. Schreck also offers advice on selecting and implementing virtualization management tools.
BIOGRAPHY: Schreck, a principal analyst at Cambridge, Mass.-based Forrester, works with IT infrastructure and operations professionals. He researches networking, storage and processing technologies and their effects on business. Current topics he covers include servers, data center automation and data center design. His areas of expertise include technical architecture planning, scale-out computing and product marketing. Schreck previously worked at 3Com Corp., where he was business manager for the company's line of wide area network and virtual private network solutions.
Download for later:
- Internet Explorer: Right Click > Save Target As
- Firefox: Right Click > Save Link As
Read the full transcript from this podcast below:
Virtualization management tools: Ready for prime time?
Karen Guglielmo: Hello my name is Karen Guglielmo, the executive editor for SearchCIO.com, and I'd like to welcome you to today's expert podcast on virtualization management tools. I'd like to first welcome today's speaker, Galen Schreck principle analyst with Forrester Research. At Forrester, Galen serves IT infrastructure and operations professionals. He researches networking, storage and processing technologies and their effects on business. Current topics include servers, data center automation and data center design. Galen's areas of expertise include technical architecture planning, scale-out computing and product marketing. Galen came to Forrester from 3Com where he was a business manager for 3Com's line of WAN and VPN solutions. His background also includes technology and healthcare consulting for Fortune 500 clients. Welcome Galen.
Galen Schreck: Hi Karen, thanks for having me today.
Karen Guglielmo: Great, and as I mentioned earlier, we're here today to talk about virtualization management tools. I'll spend the next ten plus minutes asking Galen to answer a number of questions about today's topic. So let's get started. So my first question today is: I've heard that visibility into performance, guaranteeing performance, availability and even capacity planning are some of the biggest management issues that CIO's are facing when using virtualization. Can you talk a little bit about each of them and why they are causing such a concern for CIO's?
Galen Schreck: I think we've gone beyond some of the initial concern over implementing virtualization. People are moving onto more complex issues as they try to ramp up their virtualized environments. I think if you're having just five or six virtual machines per server, maybe these things wouldn't really matter to you right? It would be a much simpler environment. But as you start running 20 or 30 virtual machines per server, maybe hundreds of virtual machines overall; dealing with that overall environment, seeing what's going on, being able to plan for capacity increases. It becomes more challenging.
So I mean, certainly, visibility into performance is a key issue. A lot of folks have system management tools that don't... I'll describe them as Legacy system management tools that don't really understand virtual server performance, because they can't see into the virtual machine and they can't really understand the health of the application that's running there. They can see, perhaps, an aggregate of the heath of the virtual server overall, when they can see if it's up or down. But when it comes to understanding if a particular application is suffering from, you know, let's say kind of a performance brown out, are end users seeing a slow down? You need more specialized tools to be able to figure out if that's happening.
And I would say the same would go for sort of guaranteeing performance and availability, too. You now have to deal with a lot of machines, virtual machines, on one box. And so people are concerned, I think, over having a lot of eggs in one basket. You've got 30 machines perhaps running on this VMWare ESX host and if something bad were to happen to it, you're gonna lose more than one app now, right? It's gonna be a sizable outage for a bunch of different customers using a bunch of different applications. So being able to see those things happening or anticipate those things happening before things get bad, and respond to them by re-locating virtual machines or spreading out the work loads. I think that that's another key capability that people are looking for. And then as far as capacity planning goes, you'd like to be able to run this environment as efficiently as you can.
We don't want to repeat the same mistakes that we made already in terms of over-provisioning servers. We have today many physical servers run, whatever, ten to 15 percent utilization. And well, clearly that's no good. And now we're seeing people fall into the trap of doing that with their virtual servers. You buy a $20,000 server, you put ten virtual machines on it, and still it's running below 50 percent of what it's capable of.
So again, understanding how much capacity you have, how much you've promised essentially, to those applications that run on there. And being able to add new servers at just the right time. Because now we're also purchasing sometimes more expensive servers than we once did. Instead of buying a $5,000 pizza box you might be buying a $20,000 HP four socket server or something. So you know, the investments are a little chunkier sometimes. So being able to anticipate that and optimize that is key.
Karen Guglielmo: Okay. And so can you mention some of the traditional systems management vendors and start-ups out there that are offering some of these virtualization management tools?
Galen Schreck: Yeah sure. So all the major systems management players, you know, HP with OpenView or IBM with Tivoli and now BMC. They all have some kind of virtualization management capability being added to different parts of their product stack. So if you're already a customer of some of those folks, there are like I say, configuration management tools or extensions you could buy to your existing platforms that may help with that.
I would argue that some of the smaller companies that are out there now, that are specializing in virtualization management, have had perhaps deeper expertise for VMWare. So for example, oh I don't know this company Quest with a product called Foglight. Or Serba has a capacity management platform for optimizing virtual machines. VKernal makes a capacity planning tool, for example. So I think they all have sort of zeroed in on virtualization and kind of provide a lot of insight into what's happening inside of VMWare and are now adding other platforms like Microsoft Hyper-V or Citrix XenServer to that.
Karen Guglielmo: Any advice you can give to CIO's on selecting the right tools to meet their virtualization management needs?
Galen Schreck: I think the key is to not go too crazy with this. We don't want too many point products. There are a lot of really interesting products out there. And I think that the key is to find something that plugs in well with your existing tools. We don't want to create management silos. We don't want to, like I said; overinvest in a lot of point products.
There are some very interesting tools out there. But I guess, thing one, is make sure that there's a specific need that you're trying to address. I don't think that these are things you need to acquire day one as you're rolling out virtualization. I think that a lot of the tools we just talked about for capacity planning or optimization are things that start to matter, like I said, when you get into running dozens of ESX servers with possibly 100 or more virtual machines on them. It's not a day one type of requirement.
And I think it also depends on the complexity. So when you have a more complex environment with, let's say, a lot of dynamic resource scheduling going on. Or where you're using a V Motion to move resources around and configure networks dynamically. You know, now some of the tools like, I don't know, HP has a thing called Insight Dynamics. And there is another company, I think it's Replicate Networks that can optimize VM placement and understand the network configuration issues that sort of surround your virtual and physical network environment.
Like I say, those things start to matter when the environment becomes more complex. So finding things that fit well with your existing management platform, be it OpenView or Tivoli or BNC, and selecting tools from providers that are what's called multi-functional. I think there are a lot of very small vendors out there. But the folks I mentioned like, oh I don't know, Serba or Quest or VKernal, their products I would argue provide a multitude of different functions. They're a little bit more Swiss Army like in nature. So you're not buying just a single feature.
Karen Guglielmo: Okay the last thing I want to ask you about is costs. Are these third party management tools really affordable to enterprise CIO's?
Galen Schreck: I think that... Well the cost varies. And I would argue that they're not super cheap. It really depends on how many sockets the server has in terms of processors. A lot of them are licensed by the number of CPU's and the number of virtual machines you have. I would argue that they are affordable in the, you know, sort of in comparison to how much you're going to be spending on the rest of the virtual infrastructure or new virtual servers and things like that. They're not terribly costly, I don't think.
And I think, again, the key thing is to not go too crazy with lots of different point products or products that have, sort of, very narrow functionality. I would think of it as just sort of extending your existing investments in system management. As opposed to trying to create a whole new silo of management. So the costs are, like I say, it really does vary. But maybe from a couple hundred dollars per CPU on up to maybe even a thousand dollars a server. It depends on the tool. But I would argue that they are, like I say, sort of comparable to existing management tool costs. So not too out of line.
Karen Guglielmo: Okay great. And on that note, that does conclude today's podcast. Thanks again to Galen Schreck for speaking with us today. And thank you all for listening. Have a great day.
Galen Schreck: Thanks very much.
This was first published in April 2009