The short answer
Server maintenance is the ongoing work that keeps a server running: patching, security updates, backups, certificate renewals and monitoring of disk, memory and load. We do it for Australian businesses from Brisbane, hosted in a Sydney data centre, with a 1 to 3 hour response.
What the agreement covers
- Operating system, package and control panel patching on a schedule
- Security updates applied promptly, not at the next quarterly review
- Disk, memory, processor and database trend monitoring, not just an uptime ping
- Certificate expiry watched separately from whatever is meant to renew it
- Backups verified and restores tested through cloud storage and backup
- Log rotation, cleanup and the housekeeping nobody schedules
- Capacity advice before you need it, including when to move to a VPS
- A named person to call, inside 1 to 3 hours
Monitoring that says you're down, and monitoring that says you're about to be
Most businesses have the first kind. A service loads your homepage every few minutes and emails you when it fails.
That's worth having, and it isn't maintenance. It's a smoke alarm that sounds once the room is alight. By the time it fires, customers have hit the error, your ads are still spending, and you're starting from a standstill.
The second kind watches slopes rather than states. Disk, memory, processor and response times recorded over weeks, so the question becomes when will this run out rather than has it.
Disk space is the most common preventable outage we see.
Not hacking, not hardware. A volume fills and everything stops. Logs that were never set to rotate, database dumps written to the same disk as the database, and the classic, a debug log switched on during a fix three months ago and left running.
When a disk hits 100% the symptoms don't look like a disk problem. The database refuses writes, forms fail, and the errors send everyone hunting through recent code changes. Knowing to check disk first turns an afternoon into ten minutes. Alerting on the trend at 70% turns it into a scheduled task nobody notices.
Certificate expiry is the other one, and it's entirely avoidable.
Automatic renewal works until it doesn't. A DNS record changes and validation fails. A new redirect or firewall rule blocks the path the renewal check uses. The certificate on the load balancer was never in the automation because someone installed it by hand. Nothing warns you, because the system that failed is the system that would have told you.
So monitor expiry from the outside, independently, with alerts at 30 and 14 days. When a certificate lapses, browsers put a full page security warning in front of every visitor and conversions go to zero.
The alert worth having fires before your customer notices.
Disk trending towards full in nine days. Memory creeping up after a deployment. A backup job that has quietly failed four nights running. Each gives a person time to act.
The other half is restraint. An alerting setup that pages for everything gets ignored within a fortnight, and the real one then arrives in a folder nobody reads.
Ask us to look at your current monitoring. We'll tell you what it isn't watching.
An alert only counts if someone acts on it
Monitoring is the easy half. Plenty of providers sell you a dashboard and leave you to decide what the red bars mean at 9pm.
Ours goes to a person. We respond in 1 to 3 hours, and the response is the fix rather than a ticket acknowledging that you noticed too. Because the same team runs your hosting, your server and often the application on it, whoever picks it up can look at the disk, the database and the code in one session.
How maintenance runs
We audit the server
What's installed, what's out of date, what's exposed, and how much room is left.
We fix the backlog
Patches, log rotation, retention rules and backups, in an agreed order.
We instrument it
Trend monitoring, certificate checks and alert thresholds tuned to your traffic.
We maintain and report
Scheduled patching, capacity advice and a person on the phone through managed support.
Who this suits, and who it doesn't
It suits businesses on a VPS or dedicated server with nobody responsible for it, and anyone whose developer has left with the server knowledge.
It's a poor fit if you're on managed shared hosting where the provider already patches everything. You'd be paying twice, and we'll say so rather than take the retainer.
Why clients hand the server over
We respond in 1 to 3 hours.
Published, checkable, and the reason most clients left an offshore host.
Australian hosting, Sydney data centre.
Onshore data and shorter round trips for Australian visitors.
The people who maintain it can read the code.
32 staff in house, so a database error doesn't get bounced back to you as a developer problem.
"Nothing has gone wrong, so why pay for this?"
True right up until the certificate lapses or the disk fills. Almost every outage we're called into after the fact was visible in the trend for weeks.