HP LeftHand P4000 Redundancy

[Cosmin Pascu]

Hello all,

I have a question about the redundancy of Lefthand solutions.

I would like to buy a HP LeftHand P4300 4.8 TB. From what i have read, the 4300 is in fact a 2-node configuration, where each node is a x86 server.

Is this configuration fully redundant?

In the quickspecs it says "Dual redundant, active-active storage controllers". Are there 2 controllers/node or does each node have its single controller? If a controller fails on one node, will i be able to access the data through the other controller?

Given that the nodes are x86 servers, what if something fails on one of the nodes (motherboard/CPU/etc)? Will i still be able to access the data stored on its hdds?


Thank you,
Cosmin

[Niels Vejrup Pedersen · Accepted solution]

Hello,

Some facts:
The HP LeftHand devices is in fact at HP DL185 G5 server with SAN/IQ installed on it.

There a a few things around the redundancy you need to know.

1) Each node has its own redundancy on the storage level - e.g. the P400 provides local RAID setup.

2) For full redundancy you will configure these two nodes as a cluster, and configure Network RAID to distribute a mirror of the data on both nodes.

Regards,

[Jason.Wilkening]

The answer is it depends. :) Each module is a controller, and when yoi cluster 2 of them you get 2 controllers. Each module you add gives you an extra controller. For redundancy you choose that per volume. You can have a volume with 2-way replication which means any block written to module a is also written to module b. You can also have a volume with no replication which is only protected by the onboard raid.

[Steven Clementi]

To add further to this.. usually a single node is configured with 2 disk arrays each with 6 of the disks in the server. This allows for 2 RAID5 logical drives per node further increasing your protection level.

You can have up to 2 drives fail on a node and as long as they are not from the same array, you won't lose any data.

Since the node is a proliant server with a Smart Array, replacing the whole server would probably not even cause any data lose as long as you have all the original drives and nothing is wrong with them.



Steven

[George Streeter]

Just know if you mirror the nodes that 4.8T of raw space will end up being about 1.8T of usable space.