VNX 5600 distribute disk drive across the DPE and DAEs

fast cache best practices

https://support.emc.com/kb/82823

https://emc--c.na5.visual.force.com/apex/KB_BreakFix_1?id=kA1700000000MSZ

General best practices

https://www.emc.com/collateral/software/white-papers/h10938-vnx-best-practices-wp.pdf

in addition could you please advise me how to configure 7 FAST VP disks to create one pool contains FAST VP and 43 SAS disks.

Let's assume the following scenario, Please consider the following distribution for tier construction :

- Create a new Storage Pool.

- Extreme Performance FAST VP tier : 4+1 RAID 5 ( HOW THIS WILL BE CONFIGURED ACROSS 7 DISKS)

- Performance SAS tier: 4+1 or 8+1 RAID 5. ( 2 hot spares and 41 remained then how to configure 4+1 over 41 or we may consider 3 hot spares and 40 remained then we can apply 4+1 RAID 5)

your help is really appreciated.

Thank you for your prompt reply.

I was reading those best practices articles but i am still confused about the FAST Cache Disks.

one of those articles describes the following general rule for Fast Cache:

As a general rule, when configuring FAST Cache, make sure all of the drives involved are either entirely contained within

enclosure 0_0 (only recommended for 2 or 4 drive FAST Cache configurations) or entirely in enclosures other than 0_0.

Either of these two recommended configurations will not result in unnecessary rebuilds on power downs.

In case the two FAST cache drives are placed in enclosure other than 0_0 , let's assume they are entirely contained within 1_0. this enclosure is not backed up by the SPS so what happens when the array powers down for any reason.

is there any uncommitted data in the FAST Cache Drives needs to written back to the intended back end disks.

My question (if you did not get me above) is how to protect FAST Cache drives in case they are split on an entire enclosure rather than 0_0 and that enclosure are not protected by SPS during power failure .

Thanks in advance,

Given R5 and max 4+1 – how would you distribute 7 disks ?

Bonus question – how with R5 and max 8+1 ?

Actually I don't know if i am correct or not however the order comes with 7 FAST VP Disks. i am wondering how can we configure a storage pool with these disk and what is RAID level should be used.

what the best practice document is mentioning that RAID 5 prefers a count of 4+1 for the best performance and 8+1 provides capacity utilization.

as we have only 7 disks then 8+1 is not an option and therefore 4+1 is the only available option (Please correct me if i am wrong).

another consideration is to avoid a drive count that result in a small remainder when divided  by the preferred drive count.

in my case 7/(4+1) give us 2 as remainder disk and R5 need 3 disks at minimum.

is there any option in VNX to configure FAST tier with R5 and 2+1 of drive count(non standard private RAID group) .so 7 disks will be configured as following:

R5 2+1 =3

R5 2+1=3

1 disk as hot spare.

I wasn't suggesting spares are part of pools, I was suggesting allocations based on what he has got and what spares would be available as a result of the config. If purchasing more drives is an option, then sure extend the drive groupings.

If he uses all 7 FAST VP efd's in the pool, with an odd grouping, he'll have no spares whatsoever for any of the EFD types in the array.

IMO, it's better to have a couple of spares than have an odd group and no spares at all. On balance, with what he has, that would be my choice.

Sure 8+1R5 is an option

if you use that profile and add 7 drives it would create a 6+1R5

Similar a 5+1R5 if you add 6 drives

BTW spares aren’t part of the pool

I would suggest to revisit the white paper about pools

Adding drives to a pool will do what makes sense within the profile that you set (raid level + preferred count per tier)

If the remainder is too small for another RG then it will create the RG before with one drive less

Example: adding 7 drives when specified R5/4+1 preferred will get you

1x 3+1R5

1x 2+1R5

Nothing else would make sense

The thing with best practices is that they aren’t hard config rules

Especially with smaller systems you cant always implement all of them to the letter

Would it be better if you had multiples of 5 for the flash drives? yes

Is it a judgement call whether to try get more disks, have some unused or stray from the BPP? yes

the fast cache white paper says that Fast Cache is non-volatile and can withstand a power loss. I am trying to remember how that was accomplished. I think it was that data in Fast Cache was also in DRAM Cache and in case of power failure would be gracefully de-staged to disk first. Brett/Rainer ...do you remember ?

it could be like the same manner used in SP cache in order to commit data to disk as it arrives. with write-through cache data is placed in the cache and immediately committed to disk and an acknowledgment is sent to initiator. 

Fast Cache is non-volatile because it resides on flash drives

Sure – its a judegement call

Personally on a small system I would prefer 51R1 to 2x21R5 for 6 flash drives in the pool.