Server Lifecycle - How does the industry consider lifecycle of server hardware? How long is too long? Server hardware lifecycle: Exploring an optimized hardware lifecycle process could be a significant cost savings for us. But how long is too long to keep server hardware? Should we retain servers until end of support? Or will advances in either technology (positive) or increased failure rate (negative) push us to more frequent replacements? So, how does the industry think of server hardware lifecycle management?
Sort by:
Usually its 5 years for servers and maximum 7 years for storages.
According the to I.R.S., servers have a useful lift of 7-years. But we generally go with 5-years as that seems to work well with OS upgrades.
Industry best practices emphasize the importance of planning and managing the server hardware lifecycle effectively. This includes evaluating technology trends, assessing performance requirements, budgeting for replacement cycles, and aligning hardware refresh with organizational needs.
We used a rotational strategy. The newest hardware went to the most compute intensive efforts. At 3-5 years it would be migrate to more basic use or maybe into a development environment. We usually retired at 8 years but occasionally held it longer. We are now 100% cloud or outsourced data center for our main data centers. We will see how we evolve the strategy for servers that still need to be physically at certain remote locations and manufacturing locations .
Infrastructure hardware retention times are dependent upon several factors.
If the hardware is fully functional then there is a question about the efficiency of the hardware for the business’ current and future initiatives. An older server may run an email program fine but if the direction is to go to cloud email, then taking that server out of service to bring in current web application firewall may make sense.
Power consumption is an important consideration. Last generation servers required more power and generated more heat, requiring more cooling in the data center. Newer servers tend to be more frugal with electricity.
Compatibility with current components is a concern. A five-year-old server may be able to take advantage of everything that was in existence three or four years ago but new technologies or different use cases available due to cost may drive the replacement decision. Not long-ago computers used magnetic hard drives then SATA SSDs began taking over the server room. Now solid state NVME drives promise Input/output operations per second (IOPS) nearly ten times higher than magnetic hard drives, using less than half the power while generating less heat. Depending upon the RAID (logical disk container) servers usually have at least four drives, which many having considerably more. A basic configuration would be Raid 0 for the OS and Raid 5 for the data. Raid 0 requires at least two drives, and Raid 5 uses four or more drives.
The accounting treatment can matter. Depending upon the depreciation schedule and the associated accounting schedules that apply it might make sense to hold or pull equipment out of service. Is the equipment considered capital, an expense or something else. Expenses are usually singular events as opposed to depreciated over time. There are accounting rules regarding computer equipment, but at a very high-level low-cost items like cell phones are usually expensed and high dollar items like enterprise virtualization servers are usually depreciated. Speak with the CFO about asset disposal.
The difference between regular and extended support can matter. Equipment is within the regular support window for a period of several years, then it falls into extended support, and then maybe security patching only. Equipment support can be around 20% of the purchase price and extended support could be twice as expensive as regular support. There are limited options for extended support and the likelihood of (partial or catastrophic) failure is higher for old “more seasoned” equipment. New parts for seasoned equipment can be extraordinarily expensive if they exist. Changing to new equipment within a plan under controlled conditions is more pleasant than replacing a production server that expired due to the stress of processing a high volume of critical business operations.