The hardware refresh cycle. It’s one of IT’s most challenging conundrums. How can you make the best use of already-paid-for server hardware while understanding the increasing application performance? Not to mention information storage and data processing requirements that drive the need to upgrade to new servers. Refresh too early and you’re throwing money out the window. Upgrade too late and you risk your company’s IT systems – and ultimately the business itself – grinding to a halt.
IT departments typically have been urged to follow a hardware refresh cycle of three to five years, but that’s best understood as a rough guide. In many cases, server hardware can be used for much longer. Particularly if the applications it supports are running smoothly. In other cases, even three years can be too slow if application requirements – big data processing is one good example – are outracing what the server hardware can handle.
In the end, no one server hardware refresh cycle fits all. IT departments must customize their refresh cycles to suit their IT and business needs. Here are 10 factors to consider when making this tough decision:
A regular IT audit is common practice in most enterprises these days, but don’t assume your team’s audit is up-to-date. Be sure you have clearly cataloged the processor power, memory, I/O and bus speeds of your server hardware. Only then can you get a feel for how much additional server power and capability you’ll need and begin to calculate the costs. A well-planned audit also can help you identify areas where you can keep older servers in service, relying on updates, maintenance or virtualization (more on that below) to extend the life of hardware versus a full swap-out.
What are you hoping to gain by upgrading server hardware? Improved reliability and availability? Support for more advanced applications? Better cost-per-performance economics? There must be a compelling IT and business reason to do a major server hardware refresh – certainly more pressing than simply pages turning on the calendar. Having a keen understanding of the needs of the business and your current IT infrastructure’s ability to meet them is crucial.
It may seem obvious that making an existing server last longer will save money. While that’s often the case, there are exceptions. In particular, out-of-warranty hardware may require extended or third-party maintenance contracts, parts and expertise – costs that must be taken into account when assessing the true costs of a server refresh decision.
A new processor launch or server architecture can have a big impact on when you should be upgrading your servers. Go in too early and you may be helping your vendor fix its bugs; go in too late and your server farm performance may lag your rivals, hurting competitiveness. Like many things in life, server hardware refresh cycles are all about timing. Everything from upgraded applications to new-generation OSs to next-generation server architectures (such as the move from 32-bit to 64-bit or the evolution from rack-mount to blade server hardware) can all have a major impact on that timing. The growth of the server industry depends on server refresh cycles. When your vendor comes pushing new hardware, make sure you consider it on your terms and needs, not theirs.
The associated costs to run those servers – including power, cooling and networking – all hit the bottom line, and hard. New systems may actually be cheaper to cool or require a smaller footprint to house, reducing costs. Conversely, new hardware may bring with it additional power requirements or require a networking boost (for instance, supporting Gigabit Ethernet) to reach their full potential. Take into account the total cost of ownership (TCO) of running your new server hardware – not just upfront price-per-CPU.
Should you maintain your own data center, split servers in a hybrid manner between your own premises and the cloud, or outsource your servers entirely? These are important choices on their own, but they also drastically impact how an IT department views and plans their server hardware refresh cycle. In most cases, it’s going to be easier and cheaper to gain access to the latest servers via a cloud partner; it’s their business to keep their hardware on the cutting edge, and shared cloud economics can help lower operational costs. At the same time – and especially if you are planning to keep servers running beyond typical refresh cycles – owning your own hardware maybe the best way to amortize hardware costs over an extended period of time.
Virtualization can stretch server hardware significantly, allowing enterprises to run software-based virtual machines rather than spinning up new hardware to create a more elastic IT environment. More virtual servers running on fewer physical machines can also reduce energy and cooling costs. But higher virtualization workloads typically require more powerful hardware (including more memory and faster I/O and bus speeds), and some VMs today are designed to take advantage of chipset-level features, further hastening hardware upgrades.
Highly virtualized environments can quickly become complex, requiring expert monitoring to manage inter-virtual machine network traffic and identify the performance bottlenecks that more powerful server hardware and more flexible virtual machines were supposed to address in the first place.
Servers should stand at the ready to serve a business’ everyday needs. At the same time, IT must plan for the worst as well. And that includes recovering from an outage or other unexpected downtime – maintaining as close to 100% availability and business continuity as possible. More modern systems – built on the latest hardware, taking full advantage of virtualization, and delivered (or at least made redundant) via the cloud – are designed to meet the continuity and recovery needs of today’s IT-intensive businesses.
Want to build your business around understanding the whims of your customers down to the tiniest detail? The kind of big data storage, processing and management required to do so simply won’t work on last-generation servers. No one keeps a three-year-old smartphone or tablet around anymore; mobile technology is moving too quickly. The same rapid refresh cycles are required in the data center as the three megatrends of mobility, big data and the cloud force IT to look at drastically enhanced server refresh cycles.
As this list of server refresh cycle tips makes clear, yesterday’s three-to-five-year server upgrade cycle has become a thing of the past. Savvy IT departments today don’t look to the calendar to plan their server refreshes; instead, they rely on a keen understanding of their IT environments, the needs of their business, and the ebbs and tides of technology industry cycles to drive a change that is more crucial than ever.