The incident taught our team an expensive yet valuable lesson. We learned that machines (servers) are bound to fail eventually
no matter how well maintained. Don’t get me wrong, I had a budget friendly computer that lasted nearly 10 years without any issues. But like cars, if a computer is moderately used and maintained, the computer will last many years without issues. Driven and pushed hard on daily basis (like business servers) it will eventually succumb to failures.
Hard working servers enable businesses to operate by storing data, enabling emails, linking communications, running software and among other essential programs. Most servers are rarely taken off line and run 24/7 (365 days a year). It is truly a work horse that goes unseen by many. But for those in the know, they are keenly aware of issues and potential failures servers can encounter as they age. With these concerns, many businesses wonder when is it the right time to upgrade/invest in new server(s). Unlike my experience (and terrible way to find out), most businesses are not looking to upgrade when their server goes bust. Thus, it is wise to review and monitor server(s) health regularly. Rarely does a server simply ‘pop’ and die. Usually, there are some tell-tale signs that may indicate an aging and/or problematic server. Below are some of the basic symptoms/situation to consider and obviously not limited to.
Computers are electronics built with mechanical parts. Some of the newer servers do use non-moving parts such as Solid State Drive (SSD). But majority of businesses still rely on the more cost effective hard disk drives (HDD) that uses spindle and arm. If your server are using HDD, keep your ears open for audible noise. The clicking or whir sound maybe a cause for concern. If the clicking noise can be easily overhead and rising above the fan noise, it is best to assess the component and its integrity. Fan noise is normal, but the clicking/whir noise can be sign of potential faulty drive. If the server is overheating or constantly above normal temperature reading, this can also be a sign of degrading hardware component as it is struggling to keep up with the load. Heat is probably one of the top reason for premature failures with servers. Monitor the Central Processing Unit (CPU), fan speed, air circulation (vent), and server temperature regularly and keep a log of spikes. If you notice a pattern, it is time for an evaluation.
Slow and Lack of Performance
This may be an obvious sign that server may be aging or overloaded. When software or data is loading incrementally slow or constantly crashing, it may be time to assess the server more carefully. An extended amount of time is considered when a software is encountering the busy icon (hour-glass) or even failing to load altogether. Using most default built-in performance monitor, administrator can check CPU/Memory performance. If the CPU is peaking but looks more like a high plateau than sharp peaks, this is a sign that the system is being over burdened (or CPU bound). Start a report/log to track the CPU bound issue to see if it's associated with specific data usage/time/ or software. If this is recurring theme, it may be another indicator of problematic server.
Storage and Software Limitation
As data continue to fill the server, it will have direct impact on its efficiency to process, retrieve, and deliver the data. Shrinking storage will cause interruption in the server as it struggles to effectively organize and optimize the data processing. Defragmentation may help, but after performing a defragmentation if the improvement is minimal, it may be a dire sign of storage space limitation. Another area of concern is with newer software installation. Newer software demand more CPU power, memory, and storage to operate effectively. Some software may not be properly compatible with existing CPU and specs. If your business requires newer software and higher recommended hardware, this may be another meter to consider.
This is more like the car analogy I shared earlier. Just like new cars, most business servers are bundled with 3 to 5 years manufacturer warranty. If the server is nearing the warranty expiration, assess the warranty coverage including possible cost of labor, parts, and total repair coverage. Cost of maintenance and repairs can sky rocket once warranty is over. Surprisingly, upgrading to a new server is usually more economical versus extending service contracts, repair costs, and replacement parts for legacy system. Another thing to note is that most servers are cycled every 3 to 5 years to keep up with the evolving and demanding hardware specs of newer software. This is something to consider as well if your server is close to the warranty expiration (or have already expired).
Aforementioned are basic and rather high level overview of common symptoms that your server may be aging or in need of assessment. Every business will have its own set of requirements to evaluate the upgrade consideration including infrastructure changes, new software deployment (requirements), budget, timing, and so on. But end of the day, always ensure that your business have the proper back up and migration plan in place if that D-day arrives. And if you are experiencing one or some of the symptoms mentioned, it may be time to consider taking a serious look for an upgrade. My recommendation is don’t wait for that pop!