How to Choose the Right Server Hardware for a Growing Business

Most businesses end up buying a server because the old one has become slow, has gone out of warranty or can’t run something they’ve just bought.

Getting size right is key; servers that are built for what’s needed today will likely need replacing in a couple of years, but buying for every possible future means paying for power and storage that sits idle.

Aim for what you expect to need in three to five years, and pick a model you can add to rather than swap out. Below are some tips on finding the right server for your business goals.

Work Out What It Needs to Run

Make a list of the programs the server will host and roughly how many people will be using each one at the same time. Shared drives and accounting packages run well on modest machines, whereas a customer database alongside a few virtual machines and nightly backups needs a lot more.

The software supplier will give you a minimum and a recommended specification, and you should go by the recommended figure. This is because the minimum only tells you how the software will start and says nothing about how it behaves once a full office is using it.

Memory, Processor and Drives

Which of these matters most depends on the job. Databases and virtual machines tend to use up memory first; file storage is mostly about how many drives you have and how quick they are, and the processor only becomes the limit with heavy calculation or video work. Memory is cheap and easy to add later as long as the server has free slots, so check that before you buy.

Drives are what fail most often in a server, which is why they’re normally set up so the same data is stored on more than one at a time and the machine keeps going if one dies. That protects you from a hardware fault but does nothing if files are deleted, encrypted by ransomware or lost in a fire, so you still need a separate backup kept somewhere else. A second power supply follows the same logic, since it adds a small amount to the price and stops the whole server from going down because one part gave up.

Tower or Rack?

A tower looks like a large desktop PC and is fine for a single server in an ordinary office. Rack servers are flat units that slot into a cabinet, and they make more sense when you’re running several because the cabling is tidier and the cooling is easier to manage. If you can already guess you’ll have a cabinet within a few years, go for rack now, since switching later usually means a weekend of downtime.

Licences, Warranty and Running Costs

The price on the quote is often a smaller part of the total than people expect. Windows Server and a lot of business software are licensed per processor core or per user, so a server with more cores than you need can cost you more in licences than you saved on the hardware.

Electricity also adds up over several years, and so does the warranty, because a cheap server with only a year of cover looks fine until something breaks and you spend three days waiting for a part. Look at what response time the warranty is promising and whether it covers both parts and labour.

Server or Cloud

Cloud hosting works well when your workload goes up and down a lot, since you only pay for what you use. If the workload is steady and the amount of data is large, owning the hardware generally works out cheaper over a few years, and a poor internet connection can rule the cloud out anyway for anything people use all day.

Plenty of businesses end up with a bit of both, keeping the main systems in the office and using the cloud for their email and offsite backups.

Choosing a Brand

A handful of servers is easier to look after when they all come from the same maker, because the management tools, spare parts and support contracts stay the same. HPE is a popular choice with small and mid-sized firms. Whatever you shortlist, confirm it takes more memory and more drives than you need on day one, since that spare room is what stops you from having to buy again in two years.