Regions
Nines runs probes from multiple geographic locations so you can detect regional outages, routing failures, and location-specific latency degradation.
Available regions
| Code | Location |
|---|---|
ord | Chicago, US |
iad | Ashburn, US |
lax | Los Angeles, US |
gru | São Paulo, BR |
lhr | London, UK |
ams | Amsterdam, NL |
fra | Frankfurt, DE |
sin | Singapore, SG |
nrt | Tokyo, JP |
syd | Sydney, AU |
How regions affect incident detection
A monitor runs its checks in parallel from every region you select. Nines uses a quorum rule: an incident is opened only when all selected regions report a failure simultaneously. A single-region failure is treated as a transient regional network problem and does not page you.
This means selecting more regions makes your monitors more conservative — they fire only on genuine global outages — and reduces false positives from one-off regional disruptions. We recommend selecting at least two regions for any production monitor.
Latency considerations
Probe round-trip time is dominated by the physical distance between the region and your
server. A probe from syd (Sydney) to a US-East server will naturally report
higher response times than a probe from iad (Ashburn). This is expected and
does not indicate a problem with your service.
When interpreting the per-region latency breakdown on the monitor detail page, compare each region's baseline against its own historical trend rather than against other regions. A spike within a region is more informative than the absolute value.
Excluding regions from the latency SLO
If a geographically distant region consistently produces high round-trip times that would drain your latency error budget — even when your service is healthy for your actual users — you can exclude that region from the latency SLO calculation.
This feature is configured per-monitor on the monitor edit form under SLO Settings → Latency SLO → Exclude from latency SLO. Excluded regions still contribute to uptime checks and incident detection; only their response times are removed from the p95 aggregate and burn-rate math.
On the latency panel, excluded regions are shown with a visual indicator so you can still inspect their raw p95 values even though they do not count toward the budget.
See Latency SLO for details on how the budget and burn rate are calculated.
Geographic coverage
The current set of regions covers North America, South America, Western Europe, and Asia-Pacific. This distribution is designed to give you a representative global view: US East and West coasts, EU, and major APAC hubs. You do not need to select every region — pick the ones that most closely represent where your users are.