Hight latency is shown on a clone and on c60

[humble magnet]

We are experiencing slow performance during an Oracle database restoration and have identified a significant latency FlexClone (not on it's parent volume) on an new deployed AFF C60 system. The latency is consistently high, around 35 ms.

According to ActiveIQ Unified Manager (AIQ-UM), the primary contributors to the clone's latency are on Cluster Interconnect, and by Read and "other" operations.
The cluster is connected via a 2x100Gb network, which should be more than sufficient.

My Questions:

1. What scenarios would cause high "cluster interconnect" latency ?
2. What might the "other" category indicate in this context?

Any insights into the underlying I/O patterns or architecture that explain this discrepancy would be greatly appreciated.

[granite cove]

any flexgroups?

[humble magnet]

No flexgrop. The LIF is located on the other node than volume. But, the indirect access should take little extra, what can we look into further on what contributed high latency on cluster interconnect ?

[sick ore]

A few years ago we had the same issue. Problem was the HA cable

[granite cove]

yeah could be a few things. i'm assuming you've opened a case?

[humble magnet]

How did you find out it was HA cable? I didn't see such errors in event log

[sick ore]

Trail and error 🙈 we had some latency issues on vmware nfs datastores when the traffic went over the ha connect. Then we replaced the first cable, still the same and then the second. This finally resolved our issue