Monitoring
Watch your application, not your inbox
Everything we build ships instrumented from the first deploy, not as a phase bolted on at the end. Metrics, traces and logs land on grafana.vallete.com, and you get an account on it.
Pipeline health
- Messages processed
- 12,842
- +8.4%
- Success rate
- 99.98%
- +0.02%
- Error rate
- 0.02%
- -0.4%
- Avg latency
- 128 ms
- -12 ms
- Active flows
- 256
- +3
- Throughput
- 2.1 GB/min
- stable
Events over time
events / min
Errors over time
errors / min
Latency p95
milliseconds
Queue depth
messages
99.98%
Healthy
Last 24 hours
Recent events
- 11:08:21INFOorder.created accepted from storefront, correlation 8f2c41
- 11:08:19INFOinvoice.posted routed to finance, 42 ms
- 11:07:58WARNpartner-api slow response, 1240 ms, retrying
- 11:07:44INFOstock.updated fanned out to 3 consumers
- 11:07:12ERRORschema mismatch on legacy feed, moved to dead letter
- 11:06:50INFOdead letter replay completed, 1 message recovered
Dashboards, alert rules and runbooks are handed over with the system. They are yours.
Numbers that mean something
Throughput, latency percentiles, queue depth and error rate for each integration. Not CPU graphs. The figures that tell you whether the business is actually flowing.
Logs you can search
Structured logs with the correlation ID attached, so one failure can be traced from the request that caused it through to the message that came out the other side.
Alerts before the phone call
Thresholds agreed with you and routed to where your team already looks. The whole point of monitoring is that nobody finds out about an outage from a client.
You should not learn about an outage from your client
Every system we deliver ships with dashboards, structured logs and alerts. Not as an extra phase at the end, but as part of the first deploy.
Client accounts only. Ask us and we will set yours up.
Running something you cannot see into? That is usually where we start.
Talk to us