backlog issue in eventstream

We have seen that recently with the Capacity Metrics App.  Allegedly there are perfomance bottlenecks that result in rows being buffered as they cannot be ingested into the eventhouse.  Raises the question of the SLA for eventhouses.

Hi Yadnapathi ,

Thanks for reaching out to the Microsoft fabric community forum.

Thanks for bringing this up. A backlog building up in EventStream when writing to EventHouse often points to a performance issue somewhere along the ingestion path. This usually happens when the rate of incoming events is higher than what EventHouse can process at that moment, which leads to buffering on the stream side.

 

It’s worth taking a look at the throughput and latency metrics for both EventStream and EventHouse to see if there are any signs of throttling or delays. Sometimes, resource limits like CPU or storage bandwidth on the EventHouse side can slow things down, especially if there’s a sudden increase in data volume. If the data coming in is very bursty or irregular, that could also lead to short-term pressure on the system.

 

Upstream adjustments, like tuning the payload size or spreading out the data flow more evenly, can often help relieve the pressure and reduce backlog. If the issue persists even with stable input rates, there may be a deeper system-level bottleneck that needs to be reviewed.

Refer these Microsoft Document:
https://learn.microsoft.com/en-us/fabric/real-time-intelligence/eventhouse

Best Regards,
Tejaswi.
Community Support