TPF CACHE STORAGE should use Ready list completion stacking longest IO first to shortest

With HPF (HIGH PERFORMING FICON) and the CPU LOOP interactions with CACHED SOLID STATE STORAGE --- Generally speaking APPs on activation will (likely) incur a cache miss which is slightly longer than a CACHE HIT.

At any given time there can be a number of HITS (x) and MISSES (y) processing by storage and completed waiting on the "Ready List." A lot TRUER when operating IOPS above 64,000, 120,000 (or higher).

The READY LIST wait although short is troubling - MISSES having the longest wait (according to QMS) suggest the longest wait should always be worked first.

Working READY LIST from oldest ensures "new start UP apps" spend less time MOD Queuing.

CYDQ should list if the first Q'd IOB is a Write or READ - Helpful in telling which occurs how often ---

CYDQ0002W 15.04.42 1 RLT DASD MODULES AT OR OVER QUEUE COUNT 100 SDA MOD VSN DUPE SS USE/T SSID QUEUE* SERVTIME STATUS
2F4F 078D WW2518 078C BSS RLT/B 208C 100 606 NORMAL
END OF DISPLAY

Same for CYEM SNS- if it can tell which a write or read IO occurred last - currently it shows only the SENSE Command which isn't helpful when trying to determine which miss may have occurred last.

CYEM0099E 21.46.06 OVERRUN, CORRCTD, SSID-2064, DVC-2B08,TYPE-3390, CSW-0E00,CCW-3E,CHPID-98,SNS-04000000 08000028 00000019 13040014 0922007F 20642B01 000000E0 00000000
The above 2B01 is an IO timeout and is a request to re-drive the (read/write) IO on a different or same path - again can't tell (read/write) which IO is taking longer.

CYED needs to indicate if long-running IO is a read or a write... it's no help today.

CYED0012W 12.18.47 MODULE QUEUE FOR 29D0 MAY BE STALLED, SSID 204D, IOB 32C3EA00

Again these tools are of little help today - they don't identify the problem IO as being a READ or WRITE. Knowing helps to determine is more CACHE or CHANNEL separation required?