MSA 2050-RAID 6, Block-level data striping with double distributed parity, over multiple enclosures

[timocoz]

Good afternoon HPE Community,

Regarding the MSA 2050 Best Practices doc, it states the following: 

“Create disk groups across disk enclosures....HPE recommends striping disk groups across shelf enclosures to enable data integrity in the event of an enclosure failure. A disk group created with RAID 1, 10, 5, or 6 can sustain one or more disk enclosure failures without loss of data depending on RAID type. Disk group configuration should consider MSA drive sparing methods such as global and dynamic sparing.”  

What I can't find anywhere ( (CLI, SMU guides, Siesmic, etc.) are specifics.  Client needs to know how many enclosures they can span a RAID 6 group across.  I can't find any limits/maximums regarding enclosures.  Could one put the maximum RAID 6 disk group of 10 drives across 5 enclosures (2 drives in each)?

Thanks in advance for any guidance, Tim.

P.S. this is an oppty, not currently installed.

[SUBHAJIT KHANBARMAN_1]

I hope the following MSA2050 Quickspec document will help you,

https://h20195.www2.hpe.com/v2/getdocument.aspx?docname=a00008276enw

 It is possible to scale an MSA up to a total of 8 enclosures including the array head itself.  This means connecting to it up to seven LFF or SFF enclosures or even a mix of the two, so to offer a total of either 96 LFF drives, or 192 SFF drives.

 Note that as the MSA 2050 SFF enclosure have 24 drive slots rather than the 25 of the D2700, that the total drive count is currently 192 rather than 199 drives.  However, if upgrading an existing configuration utilizing D2700 enclosures, then 199 drives are still supported.

Hope this helps!
Regards
Subhajit

I am an HPE employee

If you feel this was helpful please click the KUDOS! thumb below!

***********************************************************************************

[RJEA]

Hello, 

I have a similar scenario with 1 Array Enclosure + 3 Drive Enclosure, 8 SSD + 48 FC and the best practice to distribute the disks among all the enclosures as well as the creation of disk groups where there is a R1 = SSD / R6 = FC.

For example distribute E1=8SSD+12FC / E2=12FC / E3=12FC / E4+12FC and the DG creation for example:

dgA_SSD-01 Performance (RAID 5 = 4HD) E1_Slot 1-4 dgA_10k-01 Standard (RAID 6 = 12 HD /[10+2]) E1_Slot 9-20 dgA_10k-02 Standard (RAID 6 = 12 HD /[10+2]) E2_Slot 1-12 dgB_SSD-01 Performance (RAID 5 = 4HD) E1_Slot 5-8 dgB_10k-01 Standard (RAID 6 = 12 HD /[10+2]) E3_Slot 1-12 dgB_10k-02 Standard (RAID 6 = 12 HD /[10+2]) E4_Slot 1-12

[JonPaul]

Hi Tim,
There are a few moving parts with respect to the Best Practices suggestion.
1. You have multiple enclosures
2. You are using Fault-Tolerant or Reverse Cabling
3. You can create a RAID set with only the fault tolerant number of drives in each enclosure
4. You can make sure that any disk-group reconstructions happen in the same enclosure
The goal is to be able to sustain an entire enclosure going offline and your data staying online.
For example if you have 2 enclosures and use RAID 1 or RAID 10 and have the mirrors on different enclosures then if enclosure 2 should go completely away your data is still intact and accessible (if enclosure 1 disappears host connect goes away)
Another example is RAID 6 disk groups where each enclosure has 1 or 2 drives from the RAID set.  DG1 => Disks 1.1, 1.2, 2.1, 2.2, 3.1, 3.2, 4.1, 4.2, 5.1, 5.2   if enclosure 3 should lose power DG1 is CRITICAL with 2 missing disks but is online and data is accessible.  But if DG1 => Disks 1.1, 1.2, 3.1, 3.2, 3.3, 3.4, 4.1, 4.2, 5.1, 5.2  if enclosure 3 loses power then DG1 is Quarantined due to 4 missing disks and then the Pool is also offline due to DG1 being non-accessible.
The last part is making sure that if a drive fault occurs then you need to make sure that the reconstruct happened within the same enclosure: eg. Disk 3.2 is removed and Disk 3.12 is used for reconstruct.