Host list
Which machines matter, and which checks would only create noise.
Infrastructure Monitoring & Alerting
Host, service and certificate checks — with a named person who receives the alert.
The monitoring job
Visibility is only useful if an owner can act.
Checks and a named alert path. Management is how recovery happens. Not a monitoring-SaaS reseller page.
- HostCPU, memory, disk, IO, network
- ServicesWeb, database, containers, processes
- ChecksThresholds you approve
- AlertA named path, not a void
- EngineerActs if the agreement covers it
Typical checks
What we usually wire, then trim to the estate
CPU — wired to hosts you run, then trimmed so alerts have an owner.
Process
Thresholds
What “needs a person” means for disk, certificates and service health.
Route
Your team, ADGLOB under management, or both.
Review
Silence alerts nobody acts on. Add the ones that should have fired last time.
Use cases
- A production VPS that currently has no checks at all
- Certificate expiry that last showed up as a browser warning
- Disk growth that only becomes visible when the site stops writing
- A new origin after migration that still needs proving
Limits
- Monitoring does not replace management.
- It cannot see inside a provider outage the platform does not expose.
- It is not 24/7 response unless that response is contracted separately.
Related pages
Name the hosts that should not fail quietly
Bring the machine list, who should be woken, and whether anyone is contracted to act when the alert fires.
FAQ
Common questions
What do you monitor?
Typical checks include CPU, memory, disk, IO, network, processes, services, databases, web servers, containers and TLS certificate expiry, according to the agreed host list.
Which monitoring product do you use?
We select tooling that fits the estate. Named products are confirmed during planning rather than assumed on this page.
Who receives the alerts?
A named path you approve: your team, ADGLOB under a management agreement, or both. Alerts without an owner are not useful.
Does monitoring replace server management?
No. Monitoring tells you something needs attention. Management is how patching, diagnosis and recovery actually happen. They are usually designed together.