Hello, I'm migrating data from a EVA P6550 to a new 3PAR 8440 "all flash" (20x1,92TB SSD)
From scratch the array reserver about 4TB for itself... OK.
I've moved some VM (mixed) data to new LUNs in the 3PAR. About 55 VMs (vsphere), with file server, mail server, and other general use servers. All convertible data to 16kb block has been migrated with VMWare converter. Not file server yet, but the two DAG mailbox server of 1,5TB each yes, to 64kb block), and lots of other machines.
Capacity summary says thatPrivate base 4.201 GiB
Private snap 5 GiB
Shared 1.700 GiB
Free 43 GiB
Capacity efficiency says that
Compaction 1.8:1
Data reduction 1.5:1
Compression 1.2:1
Deduplication 1.3:1
Total 5.949 GiB
I ommit compaction, because I think it's related to thin provisioned luns, but data reduction is 1,5:1.. I asume that's the real benefit in disk with dedup+compresison (all LUNs with d+c in same CPG RAID6 8+2)
So my concern is that data reduction was 1,9:1 when I started moving general servers (several Windows versions, but majority of 2008R2 and 2012R2). ThenI started moving mailboxes and another SQL servers with bigger data, and I was expecting the dedup/compression ratio to increaseg (3TB of almost duplicated data in 64kb blocks), but it started to go down until the 1,5:1 you can see.
My first question: Is there any job related to scrubbing, garbaje or similar, that can reclaim space of unused blocks??? (if really there are many of them)
And second one: How can I get better ratios?? I converted mailboxes to 64kb because this is Microsoft recommendation for database disks (original machines where at 4kb), but maybe going to 16Kb is a better aproach.
bonus: If I convert again a mailbox from 64kb to 16kb inside the 3PAr, will this generate lots of dedup garbage if new blocks are different??? How can I clean old-unused blocks (like in 1st question
)???
If any of this doesn't make sense, probably is my english skill, heheh
BR.