An email conversation with the chief technology officer of AT&T's hosting and application services, who discusses the content of his Tuesday afternoon presentation at next week's Hosting Transformation Summit.
By Anastasia Tubanos, theWHIR.com
In advance of Tier 1 Research's fourth annual Hosting Transformation Summit, we're contacting some of the event's presenters and asking them about the content of their sessions in an effort to give readers a sense of what to expect.
September 11, 2008 -- (WEB HOST INDUSTRY REVIEW) -- With virtualization offerings becoming more and more prevalent within the industry, as evidenced through recent launches like Microsoft's Hyper-V product and HP's virtualization suite, it stands to argue that the virtualization space is maturing quite nicely and the question most businesses are thinking about now isn't "should I virtualize?" but "what is the best option for my business?"
Even so, a recent study by Penn, Schoen & Berland Associates found that even though 86 percent of technology decision makers have deployed virtualization projects, they are expecting to have virtualized only a quarter of their technology environments by 2010.
And security is certainly a sensitive issue amongst virtualization skeptics.
In his presentation titled "Is Virtualization Ready for Prime Time?" Tobias Ford, chief technology officer at AT&T's hosting and application services, will talk about some of the misconceptions around the limitations of the technology and discuss how virtualization is more than computing, the security aspects that should be considered and how it differs from cloud computing.
What exactly do you mean when you say "prime time?" Are there certain spaces where virtualization is working and others where you would still be hesitant to put it into practice?
Tobias Ford: "Prime time" relates to virtualization's ability to service a broad spectrum of enterprise requirements and service levels. Virtualization is not appropriate for organizations unwilling to commit to ITIL type service management processes.
Asset, change, capacity, patch, problem and vulnerability management processes are absolutely required to manage "virtual sprawl." X86_64 Virtualization, from VMware, Xen, et al., is not appropriate for all types of requirements. There is still a place for higher end UNIX servers.
Though there is virtualization available at the high end via domains zones, LPARs and such, the commodity X86_64 processor types continues to be aided by demand and Moore's Law such that very few loads require a fully dedicated server.
What are some of these misconceptions regarding the limitations of the technology you'll be addressing?
TB: Security is a particularly sensitive issue with clients and auditors alike.
For example, a common misconception is that a virtual solution cannot meet every PCI compliance rule for creating secure e-commerce sites, especially not in a multi-tenet environment. This is actually not the case and plays to one of virtualizations strengths, which has been around for awhile, 802.1q tagged VLANs.
Most virtualization platforms support tagged VLANs and by doing so can provide the logical separation needed both to meet regulatory rules for separation, such as the separation of web, app, and DB tiers with firewalls, and to make it easier to move aspects to a more directly connected type of a solution if performance becomes more of an issue and you want to reduce the latency of the interconnections within your solution.
Earlier this week, Symantec released the results from a survey that gives a rather interesting take on how virtualization has been complicating enterprise disaster recovery planning. What are some of the other challenges it presents for IT planning?
TB: The key issue is around managing performance. Whilst Moore's law continued unabated over the last few years, we started to care less and less about performance tuning and performance capacity management. We were always designing to the worst case scenario. This in turn meant, for most cases, we didn't have to understand the constraints on growth or troubleshooting required to solve performance problems.
This is absolutely not the case with virtualization. Now, with virtualization, the container or guest is right sized to the solution, to the typical usage pattern, and then grown as needed over time.
That said, the trend is to skimp a bit on capacity. When this happens you are solving performance problems more often then you used to. We have had to revive the lost art of performance troubleshooting.
Who should be looking for these solutions? Which specific people should be looking to virtualization for their certain problems?
TB: The constrained. Those who don't have enough power, space, capex or labor resources.
What are the key points you want attendees to take away from this discussion and how can they integrate these into their businesses?
TB: Virtualization is not for the faint of heart or the "IT cowboy." It takes a rigorous approach to manage virtualization and to control the inevitable "virtual sprawl."
