What a healthy integration looks like
A reference point for what normal looks like — so you can spot when something is off.
The short answer
A healthy integration shows event volume that reflects your trading activity, a near-zero error rate day to day, and events that reach Ok within minutes of a high-volume burst. A small number of retried events that eventually succeed is normal. Persistent failures, streams that go quiet while you know orders are being placed, and backlogs that do not clear are not.
How it works
Event volume
The volume of events on each stream should reflect your trading activity. A flash sale should show a spike in order events. An unusually quiet period during a sale is a warning sign.
Events waiting to be sent
Events sitting at New or Transformed are normal immediately after a high-volume burst: they have arrived, or been transformed, but have not reached the destination system yet. If an event has not progressed after 15–30 minutes, investigate. For what each status means, see Understanding event statuses.
Error rate
A healthy integration has a near-zero error rate day to day. A small number of retried events that eventually succeed is normal. Persistent failures are not.
What is not normal
Events consistently failing with the same error: this suggests a mapping or configuration issue.
A stream showing no activity during a period when you know orders were placed: this may mean the source system is not sending events.
A backlog of events at New or Transformed that is not clearing.
A sudden increase in the volume of error alerts.
What this means for you
Check the Control Panel dashboard at least once a day during active trading periods, looking at event volume, error rate and events still waiting to be sent. See Reading the dashboard for where each of these appears.
If an event has not progressed after 15–30 minutes, investigate.
If you see any of the signs listed under "What is not normal", go to An event hasn't arrived in the destination system or I received an alert — what now?.
You can only judge what is off if you know what normal looks like for your streams, so set a baseline when you first go live (see below).
Checking this yourself
When you first go live, note what your typical daily event volume looks like for each stream, and what your normal error rate is. This gives you a baseline to compare against when something looks wrong.
Related
Was this helpful?
Still need help? Ask the team