Delivering is not enough if nobody can operate it
A platform in production needs someone to watch it, update it and bring it back when it fails. Where an organisation has a technical team, that role is theirs — and my job is to make them self-sufficient. Where there is no IT department, nobody holds that role, and the platform degrades until something breaks.
Ongoing operations answer that precise situation: I deliver, then I operate, until you decide to take over.
Three maintenance types, three distinct purposes
- Corrective — restoring service after an incident. This is the core need when you have nobody to intervene: diagnosis, fix, recovery, then analysis so the failure does not recur.
- Preventive — stopping the incident from happening. Version upgrades, security patches, dependency updates, tested backups, saturation thresholds watched. This is what brings the corrective workload down.
- Predictive — anticipating degradation before failure. Error-budget burn, capacity forecasting, anomalous trend detection. Reserved for platforms where downtime genuinely costs money.
The level follows your business, not a catalogue
There is no one-size package. The first question is criticality: what happens if your platform is unavailable for an hour? for a day? Lost revenue, regulatory obligation, people’s safety, or mild inconvenience — the answer drives everything else.
| Level | Coverage | Recovery target | Preventive | Monitoring |
|---|---|---|---|---|
| Standard | business hours | Agreed with you | quarterly | availability |
| Enhanced | extended hours | Agreed with you | monthly | + performance |
| Critical | on-call | Agreed with you | continuous | + predictive alerts |
What never changes
Whichever level applies, three commitments stay identical — they are what separates real operations from being merely on call:
- Tested backups. A backup whose restore has never been verified is not a backup. Restores are proven, not assumed.
- Documented procedures. Every recurring task is written down so someone other than me can carry it out.
- Guaranteed reversibility. Infrastructure as code, access and backups held by you. Taking operations back must never depend on my availability or my goodwill.