backtop


Print 9 comment(s) - last by inperfectdarkn.. on Sep 23 at 9:38 PM

"More than 50 percent of new servers being installed in 2009 will host virtualization" -- TheInfoPro

Hypervisors are running amok in the server room. A survey published earlier this month by research firm TheInfoPro states that 50 percent of new servers installed this year will be virtualized, and predicts the number will climb to 80 percent within three years. One of the biggest growth areas for the technology, according to researcher Steve Hilton at Yankee Group, is in small- and medium-size businesses (SMBs), where as many as 69% of SMBs say they plan to adopt server virtualization in the next 24 months.

While SMBs may have been a little slower on the virtual machine uptake than their big-business counterparts in the past, they seem to be making up for lost time. SMBs typically have a higher concentration of virtualization in their IT operations according to Greg Schultz, senior analyst for IT research and consulting company StorageIO, and author of the recent book, “The Green and Virtual Data Center.” Enterprise-class companies may tend to have a higher count of virtualized servers overall, Schultz says, but “It’s more likely in an SMB environment to see a higher level of virtualization – in other words, 50, 60 70, maybe even up to 100 percent of their servers have been virtualized.”

In fact, if there’s such a thing as “over-virtualization,” an SMB server room is where you’re most likely to find it. Schultz says that in some cases, SMBs may actually be virtualizing to an unhealthy degree. These smaller businesses sometimes find themselves “going on a craze of trying to over-consolidate, and trying to over-optimize, to the (detriment) of performance.”

When taken to such extremes, virtualization can actually be an obstacle to increasing productivity, Schultz adds. He warns of a “dark side,” where SMBs may realize too late that “in the quest to try to drive up utilization and squeeze more and more out, the performance is impacted, the response time and quality of service, latency – things that can have an impact on the top line as well as the bottom line of productivity, (and affect) what your users and customers see.”

Schultz also cautions against the tendency among some smaller firms, caught up in the fervor to virtualize, to forget about “the basics, such as fault containment, fault isolation, basic high availability, business continuity and disaster recovery.” There is an inherent problem when small companies place too many virtual machines on a single server, he said. “All of a sudden, that one server has become a single point of failure,” he noted. “You’ve put all your proverbial eggs in one basket.”

Schultz recommends using virtualization in moderation, always making sure that there are redundancies in place to protect against outages and failures. Doubling up on storage and network adapters can also be important, he says. “If I have a bunch of servers, and by themselves they’re not requiring a lot of network activity, or a lot of storage or I/O activity, when I bunch them all together, their aggregate processing needs are going to be greater. So they’re going to need more CPU, more memory, more I/O capability.

“All those things need to be kept in perspective,” Schultz says, “such that in the quest to optimize the environment, you don’t inadvertently introduce bottlenecks or instability.”




Comments     Threshold


This article is over a month old, voting and posting comments is disabled

RE: Condescending advice
By Yawgm0th on 9/22/2009 11:35:28 AM , Rating: 2
quote:
Actually, the first sentence of his post has more to do with availability. If you have one server hosting a bunch of VMs and it fails, you're screwed big time. Ultimately, virtualization should never reduce the number of actual servers below two
I think that's the big "duh" moment OP is referring to. What IT professional worth his or her paycheck doesn't know this? It's not a secret or some zen-like management principle. It's common sense.

quote:
Virtualization reduces costs greatly, sure, but at the cost of a lower number of failure points.

Actually, a properly implemented virtualization architecture using clustering and shared storage has more failure points (more availability) per service than using phyiscal servers. There are fewer physical servers to fail, yes, but the services running on them have the potential for much greater redundancy at reduced cost.

I think you actually make a valid point about small business, though. "SMBs" are so frequently referred to in technology/business articles as some heterogeneous group. "Small" and "Medium" businesses fit in vastly different categories. Many (most?) small business get away with 1-3 servers and either a single systems admin or contracted IT services (or none!). Medium businesses usually have multiple services, a complete in-house IT staff, and all sorts of technology implementations that a small business wouldn't even consider.

Frankly, I think talking about small and medium businesses as a group just doesn't make sense in most IT-related contexts.


"If you mod me down, I will become more insightful than you can possibly imagine." -- Slashdot














botimage
Copyright 2014 DailyTech LLC. - RSS Feed | Advertise | About Us | Ethics | FAQ | Terms, Conditions & Privacy Information | Kristopher Kubicki