How to choose your small business in-house server: part one

In the mad dash to push all data and business applications into the cloud, there are plenty of organizations for whom an in-house server setup remains as an optimum choice for its own specific, business needs. There are plenty of reasons why small and emerging businesses consider in-house servers. Among them might be:

Security: Sensitive data need never leave the local network. This means that many common cyber attacks (such as man-in-the-middle attacks like remote spoofing of cloud providers’ addresses) are avoided. Additionally, organizations don’t have to trust a third-party with its information unless absolutely necessary.

Control and bespoke provision: Very specific set-ups are possible which are not available in the cloud, or are prohibitively expensive. Specialist companies often require this type of provision.

Connectivity: If the connection speed to the broader internet is slow, or the service unreliable in any way (one thinks of businesses operating in rural areas, for instance) then keeping data and applications in-house will help ensure business continuity.

Personal preference: Technical staff may prefer the granular control in-house services offer. Using cloud applications is suitable for many, but not for all.

With these possibilities in mind, here are some of the criteria you might wish to consider when deploying a server for the first time in your startup or expanding the business. While we have not reviewed individual models of servers, we have selected several instances which are widely available to show the range of products on offer designed for the smaller organization – see part two of this series.

1. Expandability.
Most PCs have around a three-year lifespan. Depending on the quality of components (which is largely determined by the manufacturer’s choices), some computers will last much longer, while others’ use of grey imports in their make-up, for example, means their design is effective for a limited shelf-life.

If your business is to last longer than a few dozens of months, sourcing hardware which will grow as your business does is essential, therefore. Serious consideration should be given to ensuring a maximum possible number of expansion slots, hard drive bays, and so forth – within the constraints of your budget.

2. Storage.
Disk storage space is not the budgetary concern it once was: substantial disk arrays are now available at relatively low cost. What is more important are the twin considerations of failover and swap-out.

As an absolute minimum, your server should be able of creating RAID 1 storage. This means that the device will have (at least) two drives, one which will mirror the other in the background (RAID stands for Redundant Array of Interchangeable Disks).

If you have the budget, consider three drives or more, which can be configured, for instance, in a RAID 5 configuration. RAID 5 specifically means that if one drive of the array fails, it can be replaced (more or less instantly) and the software or hardware controlling the RAID will replace the data on the failed device, byte for byte. You can read more about RAID options here.

Ideally, drives should be “hot-swappable” – this means that replacements can be slotted into a live system with no (or very little) service interruption. While hot-swappable hardware is more expensive, you may wish to consider the impact of a protracted downtime during which no data is available.

SSD or HDD? SSD drives are faster but are still more expensive byte-for-byte than their HDD variants. SSDs also lose their storage size very gradually over time, at a rate dictated by the number of deletions and writes they undertake. If your applications spend a lot of time changing your data, consider the faster end of the HDD market – 7,200rpm spin speeds should be a minimum on offer these days, and faster spin rates are available at a cost.

Finally, select the best drives you can afford. “Enterprise class” drives are more reliable but are more expensive.

3. Memory
Computers write data to disk continuously (not just when saving files or edits, for example). The more memory your server has, the less time it wastes writing data to disk when memory gets full. Memory chips are faster than any hard drive, so buy as much as you can afford. A Linux-based server will typically use less RAM than a Windows-based OS; 8GB or 16GB respectively should be considered as absolute minimums.

4. Operating System
If your organization is dependent on Microsoft-specific software, then Windows Server 2016 is an obvious path to take. But bear in mind that most servers in the world run some flavor of Linux, and therefore should be considered as a first choice.

Linux has a reputation for being difficult to configure and unfriendly to newly minted systems administrators. But enterprise Linux systems like Red Hat and CentOS can either be supported by paid-for packages, or users can turn to the wider open-source community for help if required.

Linux services are pretty much configurable out of the box these days, often via a GUI familiar to any desktop PC user. Plus, Linux service applications are usually happy to interface with Windows-based systems; the same cannot be truthfully vouched for, vice versa.

More than an after-thought

Your server will need a few extras to ensure data reliability and continued uptime. There are some essential purchases for which you’ll need to budget. Some of these are offered at the end of the web-based server configurators on pages of the major manufacturers but are essential.

1. Uninterruptable Power Supply (UPS).
UPSs are large batteries which charge as the server runs. When the electricity supply fluctuates and peaks/dips, the UPS will iron out the current’s anomalies to your expensive server hardware.

That reason alone should convince server owners of their worth, but additionally, in the event of a power cut, a UPS will inform the server of the situation, and supply it enough power from its batteries to allow the hardware enough time to power down gracefully. That means that there is a low chance of data corruption – a serious possibility if the plug is pulled on an operational server mid-read or write.

2. Backup facility.
Backup is absolutely essential, and according to the adage much older than computing itself: it’s a question of when things go wrong, not if they go wrong. You will lose data – and having a good backup strategy is the most important investment of time and money any data-dependent business can make.

Options are many, and advice should be sought from resellers or IT service providers. Options range from abundant external drives taken off-site physically, to tape drives, to full offsite backups made to an encrypted cloud. Whichever your eventual choice, make sure you practice your disaster recovery procedures, just as you might practice your fire drill. Die-hard sys admins will say the former is actually more important(!)

3. Security.
As well as physical security of the hardware, you need to install antivirus software (AV) and a software firewall as an absolute minimum. AV will ensure that anything reaching the server has little chance of infecting your valuable assets, and a firewall will prevent unwanted prying into and exploitation of your server’s possible weak points.

Both types of this type of security provision tend to come with at least the option of configuration through a GUI, and with the exception of high-end Juniper or Cisco firewall hardware, are pretty configurable by even a relative newcomer to networking technologies.

In the second part of this article, we showcase four models of server currently available on the market designed for small businesses and entrepreneurs’ startups.

Depending on your budget (from just a few hundred dollars), we hope to present a range of options which explore the ways that you can deploy, according to some of the criteria above.