Data centres & keeping servers alive

What’s a data centre?

In the last course we said a server is a computer that lives in a big secure building. That building is the data centre. It’s a warehouse whose entire purpose is to keep thousands of computers powered, cool, safe and connected to the internet — without ever stopping.

You never see it, and you never go there. You just rent a slice of the computers inside it. But it’s worth understanding, because the quality of the data centre is the quality of your service.

The four things a data centre obsesses over

Everything a data centre does comes down to four worries. Tap each one to see why it matters.

Redundancy — a spare of everything

Here’s the single biggest idea in this whole field: never have just one of anything important. If one part failing can stop the service, you put in a second one that takes over automatically. This is called redundancy.

• Two power feeds, so if one fails the other carries on.
• Batteries (a “UPS”) that bridge the gap for a few minutes…
• …until the backup diesel generators roar into life.
• Multiple internet connections, from different providers, down different roads.
• Cooling units in pairs, so one can be serviced while the other runs.

You’ll hear the shorthand “N+1”: “however many we need (N), plus one spare.”

“Uptime” and what “the nines” really mean

Uptime is the percentage of the year the service is actually up. Providers boast about “nines” — 99.9%, 99.99% — but the numbers only mean something when you turn them into how long you’re down. Try it:

See the jump? Each extra “9” cuts the downtime roughly tenfold — and costs more. For most business sites, “three nines” (99.9%) is a sensible bar to ask for; higher is nice but pricier.

Keeping the server healthy: watching & patching

A building keeps the hardware alive; server management keeps the software healthy. Two everyday jobs matter most:

• Monitoring — automatic watchers keep an eye on the server’s pulse (is it up? is the disk filling? is it overloaded?) and raise the alarm before a small problem becomes an outage. Like a heart-rate monitor by the bed.
• Patching / updates — software has flaws discovered over time, some of them security holes. “Patching” means regularly applying the fixes. Skip it and you’re leaving windows unlocked. This is one of the most important — and most neglected — jobs.

Backups & disaster recovery: the 3–2–1 rule

Everything above keeps the server running. Backups are for when, despite it all, something goes wrong — a disk dies, a file is deleted, ransomware strikes. The gold-standard is easy to remember, the 3–2–1 rule:

• 3 copies of your data,
• on 2 different kinds of storage,
• with 1 of them kept off-site (somewhere else entirely).

And the part everyone forgets: a backup you’ve never restored is not a backup — it’s a hope. They must be tested by actually bringing them back now and then.

If your business holds client records, invoices, documents or anything you can’t recreate, this matters enormously — for some firms, losing data isn’t an inconvenience, it’s a compliance failure.

Security: locks at every level

Security isn’t one thing; it’s layers, like a castle with a moat, walls and a locked keep. From outside in:

• Physical — the data centre itself: fences, guards, badge access, cameras. You can’t just walk in and unplug a server.
• Network — a firewall stands at the front door, only letting through the traffic that’s allowed and blocking the rest.
• Access — who can log in, and how. Strong passwords, ideally “keys” and two-factor, so a stolen password alone isn’t enough.
• Encryption — scrambling data so that even if someone got a copy, it’s gibberish without the key. (It’s exactly how sensitive things like passwords and access codes should be stored.)

Managed vs unmanaged: who mows the lawn?

When you rent a server, a crucial question is who looks after all of the above — you, or the host?

• Unmanaged — they give you the keys to an empty room and that’s it. You handle updates, security, backups, monitoring. Cheaper, but it’s your responsibility (or your developer’s).
• Managed — they look after the operating system, patching, security, backups and monitoring for you. More expensive, far less to worry about.

The promise they make is written in a Service Level Agreement (SLA) — the uptime they guarantee, how fast they’ll respond to problems, and what you get if they miss it. Always read the SLA.

Bringing it home: what to ask about our setup

You now know enough to ask the right things. If your business holds important client data, here’s a data-centre & management checklist to run through with any host:

• Where is the data centre — is it in the UK? (data-protection reasons)
• What uptime do you guarantee, and what’s in the SLA if you miss it?
• Is the power and cooling redundant (backup generators, N+1)?
• Who handles updates, patching and monitoring — you or us? (managed vs unmanaged)
• How do backups work — how often, kept off-site, and have restores been tested?
• What security is in place — physical, firewall, access control?
• Can we grow (storage and the GPU) without an outage?

A quick five-question quiz

Instant feedback on each — see what stuck.