High Cluster Interconnect latency for 256KB or larger IO size
Applies to
- ONTAP 9
- NFS
- CIFS
Issue
- High latency and/or low throughput observed.
- High latency mostly from Cluster Interconnect, which can be confirmed from CLI, AIQ PAS or any other performance monitoring tools:
cluster1::> qos statistics volume latency show -instance vol1
Workload ID Latency Network Cluster Data Disk QoS NVRAM Cloud
--------------- ------ -------- -------- -------- -------- -------- -------- ---------- ----------
vol1 1234 85.87ms 0ms 85.19ms 435.00us 0ms 0ms 0ms 0ms
- FlexGroup might have a mixed behavior:
- Constituent Volumes with direct access only has a low latency.
- Constituents Volumes with indirect access has a much high latency from Cluster Interconnect.
- Both CIFS and NFS would be impacted, this high latency is only observed when the Large IO size is equal to or greater than 256KB.