HPE Storage Users Group

I am confused about cage availability on the 7400. In the 3PAR guide it says this:

• HA CAGE means that no two members of the same RAID set can be in the same drive enclosure...This applies to drive chassis that are point-to-point connected to the nodes (no daisy chain)
• HA PORT applies only to daisy-chained drive chassis. When this level of availability is selected, no two members of the same RAID set can be in drive chassis that are dependent on one another for node connectivity

I have a 4-node 7400 with 8 additional SFF shelves and 2 LFF shelves. It is cabled according to the HP guide and all shelves are daisy-chained. So by the wording in the HP guide, that means that CAGE availability is irrelevant. It's PORT or MAG because if I have a cage failure, all the dependent cages in the chain are affected, so it's like a multiple cage failure.

Except it isn't irrelevant. There is a Green loop and a Red loop for the drive enclosures so if you lose a shelf there would always be the alternative path to the other node in the pair for the other shelves to connect through. So what exactly does PORT availability refer to? I am well confused.

With my number of nodes/shelves I am allowed to create 4+1 R5 sets with CAGE availability but I can't create any size of R5 set with PORT availability as it seems there is no way to distribute the LDs to satisfy this as all shelves are daisy chained.

Why then does the 7400 allow me to create CAGE resilient CPGs at all if HP are saying that CAGE only applies to point-to-point connected drives and what would my array have to look like to be able to create PORT available CPGs? Is that option there for 7000 series with few enough shelves to be point-to-point connected?

Don't worry about daisy chain it's only available on 3PAR F Serie.

On 3PAR F Serie, cages could be daisy chained : cabled 2 by 2 to one port of a cluster. cage 1 was cabled to node 0 and cage 2 to node 1, and of course cage 1 and 2 were double cabled. That way we could talk about ha port meaning you could loose attachement from cage 1 to node 0 and cage 2 to node 1, without loosing any data.
For example, that could be done on F400 with 16 cages and designing CPG with max R5 3+1 ha port rather than R5 7+1 ha cage.

S, T, 7xx0 and 1x000 are only concerned by ha cage or ha mag (= ha disk for 7xx0).

thanks for the clarification. HP should really make that clearer in their documentation and disable the PORT option in IMC/CLI for the series where it doesn't apply.

First, read Concept Guide.
Second, use SSH.
Third, enjoy