A few weeks ago I spoke with a customer who had recently purchased a new server. He was satisfied with the build quality and the software and the service we had provided, but was perplexed that we hadn’t suggested he upgrade sooner. I asked him to tell me more.
The man is the owner of a business made up mostly of accountants and analysts who spent a good portion of their days waiting for financial reports to run off the server. “An extra 10 minutes of wait time adds up when I’m paying these employees a substantial salary,” he told me. His words caught me off-guard. Most people who come to us at Puget Systems in search of a new server are already having issues with their current server. While some are planning ahead, many come to us only when their server has crashed and needs replacing.
But how do you know it’s time to replace your current server?
That’s what I like to dig into this week. I talk to customers before and after the sale. If they’ve recovered from a server crash, they are in a better position to share with me what went wrong and what they would do differently. Here are a few best-practices I’ve pickup up over the years.
3 Years or Older
I know some IT managers will say servers can last longer and some do. But after the third year of ownership, support costs escalate. According to IDC, starting in the fourth year, support costs increase about 40%. By year five, you’re staring at a 200% increase. Hold on to the server for seven years and support costs hit a mind boggling 400%. At that point you’re no longer a IT manager; you’re a fireman.
That means you want to replace your server in year three before the increased support costs kick in. Many of our customers purchase a new server at the 3-year mark because that’s when the warranty has expired. Extending the warranty can be very expensive, and you may find that replacement parts are not easy to find even when under warranty beyond three years.
Your server may be humming along just fine. I hope that’s the case. A server crash can bring a company to its knees, and that’s exactly the scenario you planning to avoid. Replacing a running server is always preferable to replacing one that’s crashed.
Non-Supported Operating System
Next month, Microsoft will pull the plug on support for Windows Server 2003. It was a vintage year and Server 2003 had a good run, but all good things come to an end. When Microsoft announces it will stop supporting the OS on your server it’s probably time to replace the server as well.
This kills two birds with one stone because it allows you to move to modern, better performing hardware while running a modern server OS that takes advantage of that new hardware while being fully supported by Microsoft. running an old server on an unsupported OS is like playing with fire: eventually you’re going to get burned.
Whatever company you select to build your new server should understand exactly how you plan to use it. MSPs can provide timely advice here because I often see small companies purchase too little or too much server. Or they purchase a server whose primary task is RAM intensive, but they configured a model with high-end CPUs but not enough RAM.
This one is difficult to define. Many employees will learn to work around a slow performing server by starting a job and then going out for coffee while it runs. A server that appears to be running reliably might not be running optimally, and this is where IT can help determine if a replacement is needed. Servers will run slow once performance reaches about 70% of maximum and/or drive space runs low.
I’ve heard customers says they don’t plan to replace their server until they can be guaranteed the new model will be 2x or 3x as fast as the old model. And while that might be possible in some rare instances, it’s not a likely scenario. CPUs like the Intel Xeon and fast SSDs have drastically increased server performance over past few years which translates to less down-time for employees. A 10-minute savings might not sound like much, but multiply that across a department and you might be surprised at the performance boost.
A good friend of mine is a dentist. He told me that viewing large images on his server was taking several minutes to load. We looked at replacing his hard drives, adding more RAM and a new CPU. But once we ran the numbers it made more sense to replace his aging server with a new model. Those images now take a few seconds to render, giving him more time to spend with patients.
There are many others reason you might want to upgrade your server, but these are the top 3 reason I hear from customers. There are also reasons you may want to upgrade rather than replace your server. One example would be that you’ve repurposed a server that just needs a little more storage or memory. If that server is a year or two old, that’s a reasonable option. But don’t let it nickel and dime you to death down the road.
Your servers are the backbone of your company’s IT infrastructure. It’s wise to have a replacement plan for each one so that you’re not scrambling to justify the replacement costs when a server crashes. You can help your company make wise decisions by reminding the boss of the inherent risk when servers are left to perform critical functions well past their life-cycle.