Server Snafus: Pitfalls to Avoid
Buying a new server or servers means making decisions about storage and processing power, applications to install, where the server will be located, and how it will be powered. Get it right, and you'll provide infrastructure to help your company grow and thrive. Get it wrong, and you can wind up spending thousands on technology that creates more problems than it solves.
Here are five mistakes small businesses often make when purchasing servers:
1. Thinking short-term
"The number-one thing we see companies do all the time is underestimate capacity and buy too small," says Cameron Niles, principal, Syzygy 3 an IT consulting and integration firm. "They get into a situation where the configuration meets their initial needs, but in a year or a year and a half, they're constrained."
How do you buy now for future needs? Begin by considering the technology's life cycle -- that is, how long do you want to be able to use it before it needs replacing? "For small and medium-sized companies, that's usually three to six years," Niles says. Of course, every company is different. But, Niles says, "As a general rule, a smart thing to do is to for storage needs to double every year."
This doesn't mean you have to buy everything today that you'll need for the next six years, he says. "You don't need to build the Death Star right out of the gate," Niles says. "But you do need to have a strategic plan that takes future growth into account. That's where a lot of small businesses go wrong -- they just look for the cheapest way to get it done now."
2. Thinking you can plug it in and walk away
"Sometimes people don't give a lot of thought to what people, processes, and tools they will need to keep a system running," Niles says. For instance, you should ask not only what an application can do, but who will support it in your area. If severs will be in geographically spread out locations, how will your IT staff effect repairs? Will they have to travel to the other locations, or will they be able to log on to the server and restart it remotely?
Another common mistake, he says, is for businesses to buy one server that contains a tape drive so they can back their vital data up to tape. "No one ever thinks, 'What will I do with the tapes if the server breaks?" Instead, Niles recommends a multi-disk backup system that triggers an automatic alert if any disk malfunctions. "With a good alerting procedure you can see that and get it swapped out before you lose any data."
3. Putting too much essential data on one server
"People try to save by combining features and applications on one server," Niles notes. While some combining is okay, he says, "Put too many different processes on too few servers, and an outage can take out e-mail, accounting systems and files, instead of just one of those."
On the same principle Nile recommends creating at least two partitions on each server, giving the operating system its own partition. That's because, when servers crash, some fixes require wiping and reinstalling the operating system. A separate partition can save your data and applications from getting wiped as well.
4. Not knowing your servers' power needs.
"A lot of folks today don't know how much power their servers are pulling," says C.C. Fridlin, vice president of product management for Avocent Corp., which provides IT infrastructure. Replacing an old server with a newer one can triple its power needs, he says. "You could wind up repeatedly tripping a circuit breaker, and not know why."
Larger data centers typically use an uninterruptible power source (UPS) to provide adequate power for each server, manage power spikes and dips, and provide battery power if the electricity goes off. If you have applications that simply can't go down, then a UPS might be a worthwhile investment, especially if your company is in an area where electric storms or other factors cause frequent power interruptions.
5. Not knowing your servers' environmental needs
"Small companies sometimes put servers in a janitor's closet," Fridlin notes. "There's a drain at the bottom with running water underneath, and mops -- lots of humidity. There's no ventilation -- it can feel like 120 degrees in there. Then they wonder why their equipment breaks."
Servers need to take in air no warmer than 70 or 75 degrees, generally speaking. But it's important to note where: most servers take in air at the front and blow it out the back. So it's obviously a bad idea to have the back of one server facing the front of another. In fact, Fridlin says, if you have empty spaces in a server rack, it's smart to install blanking panels -- metal plates provided by the rack manufacturer -- to discourage air flow between the front and the back of the rack.
You should also avoid over-cooling your work area. "Some companies cool it down to 58 degrees to protect their equipment," Fridlin says. Chill temperatures won't hurt servers, but may have other drawbacks. "Obviously it'll drive your power costs up," he notes. "And, if the servers don't have a dedicated air conditioning unit, your employees will complain about how cold it is. They may start plugging in space heaters under their desks." That can create a fire hazard, among other problems.
There are many challenges in buying servers, and of course this list can't cover all of them. But avoiding these sadly common mistakes is a good place to start.
MINDA ZETLIN | Columnist | Co-author, The Geek Gap
Minda Zetlin is a business technology writer and speaker, co-author of The Geek Gap, and president of the American Society of Journalists and Authors.