Primera Node High-Availabilty

[steez]

Hello, I have a question about the 4 node high-availability for which I have not found an answer in HPE materials (whitepapers, learning materials, etc..)

To my understanding nodes work in 2U pairs (a pair that consists of nodes in the bottom 2U, and a pair in upper 2U) but are not limited to that, which allows to mirror write cache to the node in other 2U section.

I understand that such event is highly unlikely, but my question is about the disk access - what happens if upper 2U pair fail simultaneously? Is Primera able to withstand such failure or it can only tolerate 2 node failure from seperate pairs (eg. simultaneously 1 node from upper pair and 1 node from bottom pair).

Same question applies to powersupplies - can it survive 2 node failure and if so does it matter which PSUs fail?

[sbhat09 · Accepted solution]

Hello @steez,

That is a very good question.
As you have rightly mentioned, such scenario is highly unlikely.
However, Primera's resilience in such scenario depends on the level of redundancy of the connections.
If your server/cage is connected to all 4 nodes, it can withstand both the nodes (in a pair) failure.
If your server/cage is connected to a single node pair, it can't withstand both nodes failure in the pair.

Regards,
Srinivas Bhat

If you feel this was helpful please click the KUDOS! thumb below!
Note: All of my comments are my own and are not any official representation of HPE.

[steez]

Hello @sbhat09 , thank you very much for the clarification!

Does the same applies to the power supplies?

[sbhat09]

Yes @steez. The same applies to the power supply as well.

Regards,
Srinivas Bhat

If you feel this was helpful please click the KUDOS! thumb below!
Note: All of my comments are my own and are not any official representation of HPE.

[steez]

hello @sbhat09, I have a similar question now regarding physical disks.

Primera uses RAID6 to protect data meaning it can withstand 2 disk failures. Does this apply to the whole array, meaning if 2 drives fail, Primera will offline all volumes?

[sbhat09]

Hello @steez 

Yes. Primera supports only RAID-6. Ideally the system can tolerate 2 disk failures simultaneously. In RAID-6, Volumes will NOT go offline in case upto 2 disk failures.

But Primera is much more advanced and resilient in this matter.
In your Primera system, the entire disk capacity will be divided into chunklets of 1gb each. And the RAID applies on the chunklets. Ex: if you have configured RAID-6 with set size of 8, the system picks 8 chunklets from different disks, apply RAID-6 on it and you will get 6gb usable space. Such blocks are called logical disks (LD). When 1-2 disks fail, the system will re-arrange the logical disks as per RAID-6 using the chunklets from the healthy disks.

Later, the system with RAID-6 can tolerate more than 2 disk failures as well if your disk configuration is well planned and storage capacity utilization is not very high.

Regards,
Srinivas Bhat

If you feel this was helpful please click the KUDOS! thumb below!
Note: All of my comments are my own and are not any official representation of HPE.

[sbhat09]

Hello @steez,

I have explained how to calculate space in 3PAR (same hods good with Primera as well) in another discussion chain.

I thought you may find it useful as well - https://community.hpe.com/t5/HPE-3PAR-StoreServ-Storage/3par-capacity/m-p/7160992/highlight/false#M7437 

Regards,
Srinivas Bhat

If you feel this was helpful please click the KUDOS! thumb below!
Note: All of my comments are my own and are not any official representation of HPE.