Monitoring Benefit
This is the Exchange Online row of Service Level Availability. It reports what proportion of the selected period Exchange Online was available, as measured by the logon tests ENow runs against your tenant.
Exchange Online is usually the workload that matters most here, because mail is the service users notice first and complain about loudest. It also tends to carry the highest check count, because its tests run frequently from every node.
For how the figure is calculated, the two surfaces it appears on, the Node Type filter and where the thresholds live, see Configuration - Service Level Availability.
What feeds this figure
Availability is measured by logon tests — can a client authenticate and reach the mailbox. In the report's detail grid the Exchange Online failures appear under:
| Test Name | What it proves |
|---|---|
O365EWSLogonTest |
Exchange Web Services can be reached and the mailbox opened |
O365OWALogonTest |
Outlook on the web logon succeeds |
This is a narrower set than the Exchange Online check list. Autodiscover, MAPI/HTTP, ActiveSync and mail flow all have their own indicators and their own articles, but they are not what this percentage is built from. A mail flow problem can be serious and leave this figure untouched — which is the right behaviour for an availability measure, and a trap if you expect it to summarise everything.
Read this figure for was Exchange Online reachable, and the individual checks in the Exchange Online section for everything else.
How to use it
Expand the Exchange Online row on the report and read the failures. Three patterns cover most cases:
| What the detail shows | What it means |
|---|---|
| Failures confined to one Site or Server | That location, not Exchange Online. Work the probe's path to Microsoft. |
| Failures across every node type at the same timestamps | Genuinely tenant-wide. Compare against Microsoft's service health for that window. |
Only O365EWSLogonTest failing, OWA fine |
EWS specifically — throttling, an app registration or permission change, rather than an outage. |
The Elapsed Milliseconds column is worth reading even on failures. Very high values before a failure indicate timeouts — a slow path rather than a rejected one — and point at the network instead of Exchange.
A note on hybrid environments
This figure covers Exchange Online. If mailboxes are split between on-premises and cloud, an on-premises problem will not appear here — it shows in the Exchange Server checks instead. Users rarely know which side their mailbox lives on, so when a complaint does not match this figure, confirm where the affected mailbox actually is.
Common warning or error results, and potential solutions
| Result | Potential solution |
|---|---|
| Sustained 0% uptime | Usually collection rather than an outage. Confirm the logon tests are running and the test account can still sign in. |
| Uptime below the Microsoft SLA column | Not automatically a breach. Filter by Node Type first — a single unhealthy probe drags the blended figure down. |
| Failures cluster at the same time each day | Look for a scheduled job competing for the test account or the connection, rather than an Exchange fault. |
| Users report mail problems, uptime healthy | Expected. Mail flow and client protocols are measured separately — check their own indicators. |
| PoP Trend red while uptime is high | The trend shows direction, not health. A fractional slip from an excellent figure still reads red. |
Comments
0 comments
Article is closed for comments.