This is simply measuring the depth of your queues on the FA.  The long queues may be caused by a backend bottleneck and may have nothing to do with the FA CPU.  Or it could be bursts of IO, where the utilization is high for very short periods, while the average is still low.  Another thing to consider is the number of active volumes.  Not enough active volumes can create a bottleneck on the FA CPU, where the utilization is low, but the queue is high.

thaks for your help, let me explain a little bit more, this  is a SAP ECC server with 8 dedicated FA ports, and 847 metadevices 4 member each. this is a big system and the main focus was the FA port utilization,  the main concern for the customer is the bursting, as i told you the queue depth  utilization is reaching the 75%.

Best regards

Some queuing is OK.  If response times are OK, I wouldn't worry about it.

I assume the backend is OK, and not over utilized.

Hi kikoguanabakoa

Is the 75% queue depth utilization is on your peak loading time?  Does it occur on all your 8 FA ports? or only one or several ports? If only few ports reach that, there may be performance bottleneck on your BACKEND handling IO of the devices which mapped to your problematic FA ports.