THS Articles

Getting the Edge With Blade Servers

By John Wu

Blade servers have enjoyed a lot of attention from the business community because they can provide the performance of a traditional server, but in a more compact and efficient form. Most major computer companies and a host of smaller innovators have begun selling blade servers to meet the market’s appetite for these compact workhorses.

With the majority of ISPs being small businesses, the prospect of this space saving, yet powerful technology has seemed quite attractive. A universal issue for ISPs, both small and large, is having to deal with the limits on the amount of physical space available and the amount of capital you have to invest in your ISPs. When looking at a blade server purchase, it is important to understand both the obvious and subtle factors that impact your TCO — in this case, “true” cost of ownership.

In the beginning, many ISPs had several ‘home PCs’ stacked on inexpensive wire metal shelving as servers. Not only did this take a lot of space per server, but it didn’t address the heat generation issues, especially as the more servers were added later on.

Now, one refrigerator-size rack can hold more than 100 blade servers. That same space would normally suit about 25 of today’s smallest servers. This is the blade’s advantage: because of their smaller size, they take up less physical space, increasing the data center’s processing power per square foot significantly.

Blade servers entered a marketplace where many ISPs were looking to cut capital and operating expenses but still needed performance. They seemed to be the answer to so many prayers. While blades beat out the ‘home PCs’ from a cost perspective, they are not the perfect solution, in a nice, neat box. The “true” cost of ownership is influenced by many not-so-obvious factors, and any savvy IT manager today would ask the following questions:
• Heat generation and cooling capabilities — Does the server have adequate cooling technology or will you need to purchase additional components?
• Power consumption — How efficient is the server?
• Interoperability — Are you locked into a certain technology? Are you able to purchase enhancements, add-ons in the open market?
• Do-It-Yourself installation or professional installation and consulting — Is the server easy to install or do its complexities warrant additional expense to bring in someone else to complete the installation?

Heat Generation and Cooling Capabilities
In many analyst and industry reports, heating and power consumption top the list of concerns when customers are looking at the blade server option. In some cases dense blade deployment requires additional investment in power and cooling, including purchasing additional equipment.

While blades take up less physical space, blade filled server racks would create “hot spots” within data centers — a problem that the room air conditioning units could not adequately compensate for. Acquiring additional or specialized cooling solution drives up the cost for cooling, which affects operational costs.

Many vendors address this issue by recommending that customers don’t run a fully loaded rack, but it is becoming more of a pressing problem. Others, usually independent blade server vendors, have been able to be more responsive to this issue by including exhaust fans and other reliable cooling systems in their products. This allows customers to fully utilize their blades without spending additional dollars on additional cooling.