A file system has been created on a logical volume spanning multiple LUNs During a backup, high latency is observed on the LUNs but only on one LUN at a time. The MOST likely cause for the problem…

If the file system was not spread across enough spindles, latency would likely be observed across multiple LUNs or uniformly, not just one at a time.

Concatenation (spanning) combines LUNs sequentially, meaning data is written to the first LUN until it's full, then to the second, and so on. During a backup, if the backup process reads or writes sequentially across the logical volume, it will primarily hit one LUN at a time, causing high latency on that specific LUN while others remain idle or less active. This behavior aligns perfectly with observing latency on only one LUN at a time.

If other applications were using the file system, latency would likely be observed more broadly or intermittently, not specifically isolating to one LUN sequentially during a backup.

Too much I/O activity from other applications would cause general performance degradation, not specifically high latency on only one LUN at a time in a sequential fashion.