"Holy infrastructure, Batman," Robin shouted as the two heroes breached the data center door, "these servers are running at extremely low utilization rates! We need to do something fast!"
I'm no Batman, but I agree that server utilization rates are less than optimal in many (but not all) data centers. Unfortunately, I can't blame any single problem for this. Instead, a number of problems are to blame for server inefficiency:
How could we overcome these challenges? On the first issue, software vendors could provide better sizing guidance, but frankly I don't see that happening. On the second, tool vendors could deliver better capacity management tools, and certainly that is happening, albeit slowly. And on the third issue, the processor and server manufacturers could slow down the rate of change in their technology — they could jump off the Moore's Law bandwagon, but I definitely don't see that happening.
So what's the answer? Well, if you ask some of the major hardware and software vendors in the server industry, you head back into Batman and Robin territory, because you end up with a Dynamic Duel.
Vendors not only have a vision for their dynamic environments, but they also have technology underpinnings (including virtualization), and more important, they have fancy names for their visions. In the world of Hewlett-Packard, this vision is known as Adaptive Infrastructure. In the IBM world, this vision used to be referred to as On Demand but now falls under the New Enterprise Data Center umbrella. On the software vendor front, Microsoft terms its vision the Dynamic Systems Initiative (DSI), and VMware calls it Virtual Infrastructure. The list of vendors — and the names — goes on and on.
Are the vendor visions realistic and achievable? Yes, over time. Are these visions worth investigating? Yes, any solution that lets IT organizations be more agile and gain better efficiencies from their infrastructure resources is definitely worth looking into. Are these visions perfect? No. I definitely think they're hot, but read on to learn why I think they're far from perfect.
All these vendors' visions are promising, and the underlying technology looks strong, but let's face facts: All the vendors' strategies are limited to their own backyard. Microsoft's grand design applies only to Windows. VMware's strategy is constrained to VMware hypervisor technology. HP's and IBM's visions work only with HP and IBM servers, respectively.
As a result, heterogeneous data centers (those with a mixture of server vendors) will need to deploy multiple strategies — and that is both inefficient and not hot.
Sean Chandler is a computer and network consultant who has nearly 30 years of field experience. Astro, a border collie with more than 40 dog years of data processing experience, provides technical support to his master, Sean.
In a world in which new technology is introduced at alarming rates, inevitably, some of that technology ends up being useful, and some of it ends up being merely an entertaining distraction. The ASUS Eee PC falls into this latter category. The Eee PC is a small form-factor laptop running a version of Xandros Linux customized for the consumer market. My master ordered one of these laptops when they first came out in late 2007, and he was both highly impressed and highly amused by the device. Well, at least for a little while. He was thrilled by the capabilities that this low-cost ($299), diminutive laptop offered: integrated web cam, integrated wireless networking, and a decent sound subsystem. But the tiny keyboard and small screen (7 inches, with 800 x 480 pixel resolution) soon got on his nerves, and he pushed the Eee PC from the "frequently used technology" pile to the "nice try" pile.
— Astro
data-processing experience, provides technical support to his master, Sean.Links:
[1] http://systeminetwork.com/author/sean-chandler