IBM Storage Virtualize has introduced port‑level statistics for iSCSI hosts in version 9.1.2. Until now, only node‑level statistics were available for iSCSI workloads. With this enhancement, Storage Insights can now report detailed per‑port performance metrics, enabling more precise visibility and analysis of iSCSI traffic across physical ports.

What Are iSCSI Per‑Port Statistics

• iSCSI Per‑port statistics are low-level counters and state indicators associated with a specific physical ethernet port .
• We track each iSCSI login’s physical port through session and login objects, store per‑port I/O bytes in the Nn_stats file so Storage Insight can display accurate Rx/Tx statistics per physical port for iSCSI‑attached hosts.
• The lsportstats command displays statistics for each node and physical port. When the type field is Eth_iSCSI, the statistics collection is displayed per physical port for iSCSI.
• There are no changes to the existing node‑level statistics. iSCSI IO stats will continue to be reported per node for both local and failover ports, along with the new port‑level statistics.

Interface on FlashSystem CLI

Below is an example of how the new per‑port statistics appear on the FlashSystem CLI.

Fig 1: lsportstats output on FlashSystem

Interpreting the CLI output:

Fields:

Each iSCSI capable ethernet port is represented by the following fields:

1. Port_id: Identifier for each entry.
2. Type: Set to “Eth_iSCSI” to denote that it is an iSCSI port.
3. Type_id: Matches the ‘port_id’ field in ‘lsportethernet’ view.

Counters:

The IO statistics will capture the total data sent (Tx) and received (Rx) for every iSCSI‑capable physical port. The following fields will be updated to reflect these per‑port traffic counters:

1. bytes sent to hosts
2. bytes received from hosts
3. iSCSI commands sent to hosts
4. iSCSI commands received from hosts 

Note: In 9.1.2 release, slow read/write and aborted read/write counters remain inactive and will appear as zero.

Why Per‑Port Stats Matter

• Port‑level statistics help identify traffic imbalance by showing which physical ports carry disproportionately high iSCSI workloads, revealing bottlenecks or uneven distribution across the node.
• These insights enable smarter rebalancing of iSCSI logins and port‑sets, ensuring traffic is spread more evenly across available physical ports to avoid overload on any single link.
• Balanced utilization maximizes compute and network efficiency, allowing the storage system to deliver optimal performance—especially important when supporting a large number of simultaneously connected iSCSI hosts.

Interface on Storage Insights

The port-level iSCSI stats are written in the Nn_stats files which is consumed by Storage Insights and displayed through the dashboard.

Storage Insights allows users to select from a range of available performance metrics and instantly visualize the corresponding port level I/O statistics. This flexible view makes it easy to analyze per‑port activity and identify trends or imbalances, as shown in the example below:

Fig 2: Chart metrics on SI

Based on the metrics selected, Storage Insights presents the I/O statistics in a graphical format.

Fig 3: Port performance on SI

Conclusion

The addition of port‑level statistics in IBM Storage Virtualize 9.1.2 brings clearer visibility into how iSCSI workloads use physical ports. With detailed Rx/Tx counters now available in both the CLI and Storage Insights, administrators can quickly spot traffic imbalances, optimize port usage, and improve overall performance across iSCSI environments.

Discover more about port‑level statistics from Storage Insights perspective through Storage Guru.

Stay tuned as we introduce Port-level statistics for NVMe Host Attachment, along with slow read/write and aborted read/write counters for iSCSI.