Symmetrix - QFULL limit

In CLARiiON/VNX...

"A high degree of request concurrency is usually desirable, and results in good resource utilization. However, if a storage system's queues become full, it will respond with a QFULL flow control command. The FE port drivers return a QFULL status command under two conditions:

- The practical max number of concurrent host requests at the port is > 1600 (port queue limit)

- The total number of requests for a given LUN is (14 * (the number of data drives in the LUN)) + 32

The host response to a QFULL is HBA-dependent, but it typically results in a suspension of activity for more than one second. Though rare, this can have serious consequences on throughput if this happens repeatedly.

HBA queue depth settings usually eliminate the possibility of LUN generated QFULL. For instance, a RAID 5 (4+1) device would require 88 parallel request ((14*4)+32) before the port would issue QFULL. If the HBA queue-depth setting is 32, then the limit will never be reached."

So, 12,288 is a bit on the high side, is it total per director, or port by itself? Would it be 2048 per port? Can you provide the link to the doc please?

On Symmetrix VMAX, each FA CPU has it's own limit of 12,288.

Not sure what you mean by host requests?

Do you mind to share the doc?

The number is per FA CPU, not port. Per port would be half that (6,444). Yes it is a considerable number. It is most unlikely for any host to get a queue full condition from the array with the queue resources that VMAX front-end has.

Alas, not contained in VMAX documentation.

That's why I can't find the doc for reference. How do you know about this piece of information?

is it correct to say that the port limit is half of the cpu limit? I would think the absolute port maximum would be the cpu limit if the other port isn't doing anything. Or is it hard coded so the limit is actually 6144?