Slow NDMP Backups – Throughput Degraded Due to Network Latency and Packet Loss
Applies to
- NetApp ONTAP 9.12.1x
- NDMP
- Commvault
Issue
- NDMP backups from the NetApp ONTAP cluster to a MediaAgent and then to a NetApp E-series system are experiencing very slow throughput – approximately 50 GB/hour, whereas previous performance was around 900 GB/hour
- This issue affects all NDMP backups, while other Commvault backups maintain expected throughput
- NDMP logs show successful backups but low throughput
- Packet trace analysis shows high network latency (RTT ~12ms) and some packet loss between the ONTAP cluster and the MediaAgent, despite both being in the same datacenter
- ONTAP performance data (CPU, disk, system latency) does not indicate a bottleneck
- NDMP “dump to null” operation achieves ~4.7 TB/hour, confirming ONTAP is not the bottleneck, at least not read:
dmp Tue Sep 9 10:13:32 PDT 2025 /vol/testvol/(0) End (2064488MB, 1825secs, 1131MBps)dmp Tue Sep 9 10:13:38 PDT 2025 /vol/testvol/(0) Log_msg(Phase 4 throughput (MB sec): read 1325 write 1315)- Commvault reports delayed responses from the NetApp NDMP server, which aligns with observed network latency
