Reliability

The one dimension where reviews and measurement overlap, and what to do when they disagree.

Reliability is scored by reviewers, and it is also the dimension the probe network measures. The two are shown separately, and comparing them is one of the more useful things a reader can do on a provider's page.

What reviewers are asked about

• Unplanned downtime they actually experienced.
• Whether maintenance was announced, and whether the announcements were accurate.
• Data integrity: whether anything was lost, and what happened when it was.
• Whether the provider's own status reporting matched reality.

What monitoring measures instead

Reachability of the provider's announced address space, from many places, continuously. See How monitoring works.

These capture different failures. Monitoring sees the 3am twenty-minute outage that every customer slept through. Reviewers see the storage failure that lost a week of work, which monitoring cannot see at all because the network stayed up throughout.

When they disagree

Both directions happen, and both are informative.

Good uptime, poor reliability reviews. Usually means failures that are not network failures: storage, hypervisor, control panel, or maintenance handled badly. The network was reachable the whole time and the service was not.

Poor uptime, good reliability reviews. Often means the problem is regional. A provider unreachable from one part of the world and fine everywhere else has genuinely reliable service for its customers and a genuine reachability problem for everybody else. The regional breakdown is where to look.

Which to trust

Neither, alone. Monitoring is objective and narrow; reviews are broad and subjective. A provider that looks good by both measures is a much stronger signal than one that looks good by either, and the listing is arranged so you can see both without hunting.