||Add To My Personal Library
January 28, 2011
Vol.33 Issue 2|
Page(s) 12 in print issue
A Clear View Of Virtualization
Understanding The Benefits & Drawbacks Of Virtualizing Storage, Applications & More
Virtualization is a buzzword that's been tossed around so liberally over the past few years that there seems to be this assumption that everybody knows its meaning. â€œIt's a word that's become overloaded,â€ says Scott Morrison, CTO at Layer 7 Technologies (www.layer7tech.com). â€œWe've been using it a lot throughout computing, and [each time] it means something a little different.â€
Morrison boils down the fundamental definition for virtualization as creating a layer between two different views of computing technology. â€œThere is hardware, the sort of pure view of actual hardware, devices, and other physical stuff. And then the next view is the virtual view, which can essentially make better use of those underlying resources,â€ Morrison explains.
Perhaps you already knew this, or at least had a sense about the basics of virtualization. After all, some forms of it, such as virtual memory, have been around for close to 40 years. But despite virtualization's seeming ubiquity, plenty of IT professionals still have only a fuzzy idea of what virtualization does, what advantages it offers, and what complexities it brings. So consider the forthcoming information as a primer of sorts that can give you a better understanding of the concept and not make you feel as though you just walked into â€œThe Matrix.â€
Abstractions & Efficiencies
David Marshall, founder of VMBlog.com (www.vmblog.com) and an editor of â€œVirtualization for Dummies,â€ describes the 20,000-foot view of virtualization as an abstraction layer. â€œIt's no longer about a massive amount of servers or disks. You start to abstract away the underlying hardware and carve things out,â€ Marshall says. â€œIn server virtualization, you have underlying hardware chopped up multiple times into virtual machines (VMs), all of which think they are that underlying level of hardware.â€
These abstractions lead to efficiencies, says Nathan Coutinho, virtualization solutions manager at CDW (www.cdw.com). â€œOn a grand scale, virtualization is all about trying to make data centers more efficient, trying to save money through those efficiencies, and making it easier to manage data center resources,â€ Coutinho says. â€œVirtualization technology itself helps you do more with less by combining multiple [IT] technologies together.â€
Server & Storage Virtualization Basics
Server virtualization has probably had the most traction in data centers so far. â€œIts whole concept is the ability to take multiple physical machines and consolidate them onto one. Instead of running 100 physical servers in a data center, you could effectively run 100 VMs on 10 servers,â€ Coutinho says. â€œThose 10 servers might be beefier servers, but you can run the same workloads in one-tenth of the space. From an efficiency and TCO standpoint, it's a huge power and cooling savings.â€
The primary cost for server virtualization (beyond hardware refreshes, which happen every three to five years) is the cost of software. Although free virtualization software is available, they lack the management capabilities and support that paid versions offer, Coutinho says.
Storage virtualization works in much the same way. â€œIt lets you combine multiple storage units to make it look like a [single storage unit] so that you can move things around and use your storage more efficiently,â€ even if you are using physical storage devices from different vendors, Coutinho says. Storage virtualization tends to be more popular in larger enterprises because of the cost of SANs, as well as the software needed to virtualize them, he adds.
Client Virtualization Basics
â€¢Â Data centers generally focus on three types of virtualization: server, storage, and desktop.
â€¢Â Ideally, virtualizing portions of your data center will lead to improved efficiencies in physical hardware, as well as cost savings from hardware consolidation and decreased power usage.
â€¢Â Virtualization does introduce complexities and visibility issues to the data center, which makes planning and management of implemented systems necessary to realize virtualization's advantages.
Client virtualization comes in several forms, but the two that have the most relevance for the data center are application and desktop virtualization. â€œBoth let you deliver applications and desktops to end users in a centralized, efficient manner,â€ Coutinho says. For example, application virtualization lets data centers store an application centrally that end users can access via Citrix (in most cases), so that data center managers don't have to install the application on every PC.
Desktop virtualization has become trendy lately because so many data centers are rolling out Windows 7. â€œDesktop virtualization lets you effectively build a Windows 7 virtual image in the data center once and be able to blast that out to thousands of users simultaneously,â€ Coutinho says. â€œYou don't have to worry about people's laptops and desktops. You just give them a URL, they connect to a desktop, and they're done.â€
Morrison says a lot of organizations are buying into desktop virtualization because of the economies of scale and administrative efficiencies that result from leveraging it. â€œCentralized control means that [IT and data center managers] don't have to spend as much time going to individual workstations, which have been configured with all sorts of idiosyncratic stuff on them. Serving a consistent [disk] image from a central location is easier to maintain, back up, and debug,â€ Morrison says. Because the data center is powering out the desktop, an organization also saves money on the cost of client PCs, he adds.
At first glance, virtualization comes across as a win-win situation for data centers. But Marshall points out that virtualization isn't just a matter of achieving a good consolidation ratio or reducing power consumption.
â€œIn the old, physical world, an IT admin or a data center administrator could walk up to a machine and know [where] a machine is in the rack, how it's cabled, and what's on it. But with virtual machines, your environment is no longer static. Your VM could be on any rack on any [physical] server, which creates management complexities and visibility issues, especially when you need to troubleshoot,â€ Marshall says.
InIn addition, data centers can potentially lose control of the procurement process when virtual machines, particularly servers, can be created with a simple right-click and copy of an existing VM. â€œThat adds to the visibility problem because if you have 50 VMs spread throughout your infrastructure, you may lack a good way of knowing which ones haven't been updated, are no longer in use, or are not in compliance,â€ Marshall says. â€œTherefore, you need proper controls in placeâ€ to handle the changes virtualization brings about.
by Robyn Weisman
Key Terms |
Application virtualization: The The ability to push out applications to end users from the data center rather than loading those applications on each client desktop or laptop. Â Desktop virtualization: The The ability to push client desktops from the data centers that end users access via a URL. Server virtualization: The The ability to layer multiple software-based servers on top of physical hardware servers. Storage virtualization: The ability to provide storage functionality across multiple physical storage devices, even when the physical hardware being used comes from different brands.