Performance impact using SnapMirror
Applies to
- ONTAP 9
- SnapMirror
Issue
- Increased latency can be seen on 'Disk' / 'Disk-io' and/or 'Data' Delay Center (D-blade) in
qos statisticsoutput due to active SnapMirror transfers - There may also be high cpu activity / utilization seen.
- Note: Data latency during a SnapMirror transfer is rare due to being a background process on CPU
Example: The following commands show high disk latency or Data latency while a SnapMirror transfer is happening
nas1::> qos statistics volume latency show -volume data -vserver data_svm3
Workload ID Latency Network Cluster Data Disk QoS
--------------- ------ ---------- ---------- ---------- ---------- ---------- ---------
-total- - 0.00us 0.00us 0.00us 000.00us 0ms 0ms
data-wid307 307 300.00ms 0.00us 0ms 150.00ms 150.00ms 0ms
-total- - 0.00us 0.00us 0.00us 000.00us 0ms 0ms
data-wid307 307 892.00ms 0.00us 0ms 92.00ms 800.00ms 0ms
-total- - 0.00us 0.00us 0.00us 000.00us 0ms 0ms
data-wid307 307 125.00ms 0.00us 0ms 25.00ms 100.00ms 0ms
-total- - 0.00us 0.00us 0.00us 000.00us 0ms 0ms
data-wid307 307 332.00ms 0.00us 0ms 300.00ms 32.00ms 0ms
CLUSTERB::> snapmirror show
Progress
Source Destination Mirror Relationship Total Last
Path Type Path State Status Progress Healthy Updated
----------- ---- ------------ ------- -------------- --------- ------- --------
svmprod:volprod XDP svmdr:voldr Snapmirrored Transferring 27GB false 07/08 11:01:52
- If snapmirror is the primary task as part migration activty and the rate of migration is low with very low throughput.