Cosanti Foundation runs Arcosanti, an architectural project in the Arizona desert, and a public website and set of online services that a great many people rely on to find it, visit it and support it. For a little over three years we were the engineers responsible for keeping those servers patched, backed up and online.
The review the foundation left on Upwork at the end of the engagement is public: "Working with Dhaval is fantastic. He's very smart and broad with knowledge and skill. A team member!"
The problem: public services with nobody on the servers
The foundation's web presence had grown over the years into several services on several servers. They were important, they were public, and nobody on the team was a server administrator. Updates were applied when someone remembered. Backups existed in the sense that a job had once been configured. When something went down, finding out why took longer than fixing it.
What they needed was not a project. They needed someone on call who treated their servers as a standing responsibility.
What we looked after
- Hosting and server administration. The servers, the web stack and the services on them, kept configured correctly and documented.
- Security patching. Operating-system, web-server and language-runtime updates applied on a regular schedule, with the riskier ones tested before they reached production.
- Backups that were actually restorable. Automated backups of files and databases, held off the server, and restored periodically to prove they worked rather than assumed to.
- Uptime monitoring. External checks on every public service, so that an outage was known about before someone from the public reported it.
- Incident response. When something did break, the same engineer who maintained the servers was the one who answered, which is why fixes were fast.
How it worked
The arrangement was deliberately boring, which is the highest compliment a server retainer can earn:
- A monthly rhythm of patches and reviews, with anything notable reported in plain language.
- Changes rehearsed on a copy before they touched the live servers.
- A single point of contact who knew the whole setup, so a question got an answer rather than a ticket.
The outcome
Three years of a public website and its services staying online, patched and recoverable, for an organisation that never had to hire a systems administrator. The engagement closed with the five-star review quoted above.
Have servers nobody owns?
Our care plans cover hosting, patching, backups and monitoring for the Laravel applications businesses depend on, with a named engineer and a response-time SLA. See the care plans or talk to us.