Additional POOL capacity metrics

Alex_Rogov

Hello,

It was nice to see you've added pool virtual capacity graphing for Storwize arrays in v. 2.40. Hitachi boxes have similar metrics but they are sadly missing from stor2rrd. Can we fix that?

1. The metric similar to IBM's Virtual Capacity is called "Subscribed Capacity" and you can retrieve it using RAIDCOM. Example output below

raidcom get pool -key basic -I<instance #>

PID  POLS U(%)  LCNT SSCNT Available(MB) Capacity(MB) Snap_Used(MB) TL_CAP(MB) BM  TR_CAP(MB)  RCNT   Seq# Num LDEV# W(%) H(%) SFSP    VCAP(%) TYPE PM PT  POOL_NAME
000  POLN  21     69     0     126285600    161055384             0   45778656 NB           0     0 ------  64  4128   80   98 YES          99 OPEN N  THP ---
001  POLN  33     63     0      53532780     80525592             0   33829194 NB           0     0 ------  32  4224   80   98 YES          99 OPEN N  THP ---

From the RAIDCOM output, we need the TL_CAP(MB) - it is 45778656 for pool 0 and 33829194 for pool 1. Just plot it on the POOL capacity graph (total and per pool) with a thin black line and call it "Subscribed".

2. Another metric (sort of unique for Hitachi arrays) is FMC capacity. Hitachi arrays can be equipped with flash modules with built-in compression capability. The amount of space used on FMC drives can be retrieved with the command below:

raidcom get pool -key fmc -I<instance #>

PID  U(%) ACT_AV(MB) ACT_TP(MB) FR(%) FMC_LOG_USED(BLK) FMC_ACT_USED(BLK)
000    21  126285600  161055384   100       66913873056       52832720298
001    33   53532780   80525592   100       55237424064       35542519182

ACT_TP(MB) is the total capacity of FMC modules included in the pool in megabytes; FMC_LOG_USED(BLK) is the amount of capacity on the FMC drives before compression; FMC_ACT_USED(BLK) is the used capacity of FMC modules after compression. Both are in multiples of 512 bytes (BLK). You can put these metrics on a separate tab (e.g. "FMC") in the "Capacity" and "Pool" sections and call them "Total", "Raw data" and "Compressed".

Pavel

Hi Alex,

we have just enhanced virtual capacity/oversubsubscription for next bunch of storages, but not for VSPG.

I think that point 1. should not be a problem.

Let us think about 2.

we will let you know

Alex_Rogov

The added metrics are working just fine with every Hitachi storage we have (from original VSP to VSP F1500 and the newest VSP G350). Stor2rrd already exceeds the functionality of any third-party product when it comes to Hitachi storage support.