RDS volume - Latency & I/O spikes

Question

we are experiencing sudden spikes in read/write latency and queue depth on a production PostgreSQL instance.
There is no increase in connections, workloads, web layer traffic, cronjobs, etc. 
We saw the physical device read/write IOs plummet on Enhanced Monitoring.

We think there to be a degradation issue with volume.

![Enter image description here](/media/postImages/original/IMifSWH7duRJmR-sE42IqL1w)
![Enter image description here](/media/postImages/original/IMA2Wtw9PCRyqxgxWeCK7m7A)
![Enter image description here](/media/postImages/original/IMMDxDtdwASuuFo7JymXDQag)

we found that it will return to normal every 5mins in our web apps apm monitor.
![Enter image description here](/media/postImages/original/IM1X67ibjATw6TzDSEpqYZoQ)

Can someone from AWS please look into this ASAP?

Answer

Check the database, to see what is causing the spikes. Performance Insights would help.https://docs.aws.amazon.com/AmazonRDS/latest/UserGuide/USER_PerfInsights.html.  With Postgres, Vacuum can be causing unexpected I/O .

Answer

This may require other details , so we would recommend to open a support case so that it can be looked/troubleshooted in more detail. In the meantime, you can also check

1. https://aws.amazon.com/premiumsupport/knowledge-center/rds-write-latency-spikes/
2. https://aws.amazon.com/premiumsupport/knowledge-center/rds-latency-ebs-iops-bottleneck/

RDS volume - Latency & I/O spikes

Conteúdo relevante