Expansion Planning For Systems Growth

Something scary is happening in the world of computing. Even though everything is getting smaller and more powerful, we’re faced with increasing demand for power to run systems, keep them cool and house them. Although disk drive capacity requirements double every year, if companies require more drives, the issues become how to deal with the increased expense to run systems and keep servers cool — and where they will house everything.

“Server consolidation and virtual servers are a major step toward improving capacity utilization of servers,” says Arun Taneja, founder and consulting analyst at Taneja Group, a storage and server analyst consultancy in Hopkinton, Mass. “Every data center person should be looking at both. Companies are also turning to blade servers as another strategic way of consolidating space.”

Restricted Space

Companies in cities such as Manhattan and London are facing serious storage issues because there physically isn’t room for them to increase the size of their data centers. Many organizations are being forced to move their operations to locations with more open space.

Yet moving a data center is among the hardest processes from a planning and execution perspective. The volume of data that companies possess is daunting, and they don’t have the luxury of simply turning systems off and putting them onto a truck.

“It’s a very expensive proposition, even in a best case scenario,” says Steve Duplessie, founder and senior analyst at The Enterprise Strategy Group, Inc., an IT consultancy based in Milford, Mass. “All of a company’s applications can be dynamically migrated online to a new site, but IT must make sure everything is synched and online before doing a hard cut over and decommissioning the old site.”

But Duplessie cautions that such moves never go perfectly smoothly and it takes an enormous amount of time — at least a year — and bandwidth on top of the cost to complete a move and upgrade.

Starting Over

On the plus side, moving a data center enables a company to start with a clean slate. It can be an opportunity to recognize the past infrastructure mistakes and to architect and configure for the new millennium going forward.

Duplessie suggests that if an old data center site isn’t going to go away, it can then be used as a disaster recovery backup site. He believes companies are starting to recognize that assets that have depreciated are ideal for housing second tier applications, or for disaster recovery.

“Many companies prefer to get rid of their old assets and tend to buy infrastructure for infrastructure sake,” says Duplessie, “which solves one problem and creates another — that’s the nature of computing.”

Another way to conserve data center space and plan for future growth, says Duplessie, is to evaluate data and determine what is mission critical and what isn’t. “People are starting to realize that not all data should be treated exactly the same and not all data needs to be on the highest tech, most expensive gizmos.”

“I would be willing to bet there is huge amount of data that doesn’t need to be spinning around twenty-four hours a day,” he adds. “Data’s value and access patterns change over time, and because we architected it and stored it there on day one doesn’t meant that’s where it should be on day two-hundred.”

If companies took 75 percent of their data off their primary production storage area and moved it to a lower cost, lower power, lower performing storage area, the savings garnered could be doubled, according to experts.

“If it’s not possible to grow physically we have to take a data-centric view,” says Duplessie. “If we place everything around the life cycle and value and importance of the data, which is variable over time, and don’t treat it the same, we have the opportunity to make decisions based on whatever criteria we have. It’s treating everything the same that gets us in this power, cooling and space problem because it’s easier to make the decision to buy more of what we have — but it’s usually strategically worse for that operation.”

Over the long term, companies need to recognize that data has a life cycle and it can be staged on different servers that have different attributes around performance. That will let IT devise plans that capitalize on existing data center usage and minimize the need to add real estate to duplicate what is already in place.