P4000 SAN Network RAID vs Remote Copy

[Mi77]

Does it mean that for Network RAID task is for failover and redundancy only. That is, if one (RAID 5) or two (RAID 6) of the networks (local or remote) fails the other remaining networks will take over.

While Remote Copy takes care of the local or remote network's storage content's redundancy, sort of like a storage mirroring?

Is this the main difference between the 2?

Coz it sounds to me like they are just the same, both copies data from one location to the other(s) for redundancy.

[Bryan McMullan]

Remote copy is primarily used for low bandwith (or variable) bandwidth situations. We use it for copying to a colocation facility used for DR. Whereas Network RAID is for fully access, instant resiliency.

Typically, you wouldn't connect to a remote snapshot/copy in normal use as it would (again normally) have higher latency and bandwith issues (over a WAN connection or the such).

Not to say you couldn't if you had the lines, but in that situation where you have GigE connections between sites, you may want to look at Campus SAN which would allow failover between sites.

So, to get to your point.... They are both, in essence copies of data from one location, but they are used in different ways (resiliency versus backup).

[Mi77]

Hi Bryan,

So does it mean that P4000 SAN does not limit itself into Local LAN resiliency only right? It's resiliency is extended up to a remote network (WAN) correct?

That is, if one of the remote site's Storage Node failed, Network RAID will automaticall connect the operating Virtual Machines from the failed site to a working or operating site utilizing a working P4000 Node(s).

Is this the understanding?

[Bryan McMullan]

Depending on your WAN connection, you are correct you can have more than LAN resiliency (which is the Network RAID part). If you have GigE or higher connections between two sites (this can be in the same building, different buildings, etc), you can setup campus SAN and have all connections failover to a secondary site on complete failure of a primary site.

Network RAID is more LAN based. If you have a volume protected by Network RAID (2 or higher) and a node fails, all traffic is automatically redistributed among the remaining active nodes until the downed one can be repaired.

Does that make sense?

[Mi77]

Sure it does make lots of sense Sir. Thank you so much, it crystal to me now :-)

[Mi77]

Hi Bryan,

Sir, one last clarification.

Can I conlude that HP's intelligent SANIQ software triggers the use of it's built-in Remote Copy when data is replicated remotely over WAN (i.e. sites located accross different countries).

While in a high speed infrastructure like MAN or SANs located in the same physical or geographical location, SANIQ will trigger the use of it's Network RAID instead, right?

In the attached document, does it mean that if these 2 datacenters are located at 2 different countries, is it fair to say that the SANIQ software will trigger data replication using the Remote Copy?

And that, referring to the same attachment, if these 2 datacenters are seated in the same location (i.e. Campus SAN), the SANIQ software will instead utilize it's Network RAID feature, correct?

[Bryan McMullan]

Mi77,

Which method used is based on how you have your SAN designed. Typically you will break your SAN out into management groups at different locations.

For Example:

Network RAID - One location with three nodes, volumes are set with Network RAID 2. If one node fails, the SAN will continue to function without any issue. The failover is instant and handled within the SAN/iQ software itself.

Remote Copy - Two locations (A and B) with three nodes each set in two management groups. Location A is the primary, location B is the backup and they are connected with a 25Mbps WAN link. Volumes are set with Network RAID 2 and snapshots are scheduled to remote copy hourly. If one node fails in location A, Network RAID will handle failure and all connections stay at location A. If a second node fails at location A, the SAN will go down. You have backup data at location B....but you are basically down and out (unless you have a DR plan setup to use the backup data).

Campus SAN - Two locations (A and B) with three nodes each set in one management group. Both locations are connected with connected with a low latency GigE MAN Link. In the example above, if two nodes fail at location A, all connections will failover automatically to location B and continue to function. But if only one node fails, Network RAID will handle the issue and all links continue to be served at location A.

I hope that clarifies it a bit better.

In direct answer to your questions:

1. While in a high speed infrastructure like MAN or SANs located in the same physical or geographical location, SANIQ will trigger the use of it's Network RAID instead, right? This is highly dependent on your design. As a general answer though, yes if all nodes are within the same cluster.

2. In the attached document, does it mean that if these 2 datacenters are located at 2 different countries, is it fair to say that the SANIQ software will trigger data replication using the Remote Copy? I believe this is using VMWare HA, which would require high speed (GigE links or higher) and would be more like the Campus SAN solution.

3. And that, referring to the same attachment, if these 2 datacenters are seated in the same location (i.e. Campus SAN), the SANIQ software will instead utilize it's Network RAID feature, correct? Again, this is based on design. If within the SAN/iQ software, you have all nodes in the same cluster/site, then it will use Network RAID. But you could also design it with the Campus SAN feature, it highly depends on your needs and your redundancy requirements.

Please keep in mind that how things work is highly dependent on how you design it within the SAN/iQ software itself.

I hope that helps!

[teledata]

And to really boil it down for simplicity

Network RAID = a SINGLE cluster of 2 or more nodes operating as one pool of storage.
NETWORK RAID = automated high availability (when coupled with the Failover Manager) of storage over LAN or high speed (must also be low latency) WAN. Volumes are always sychronously striped over all members of the cluster


Remote Copy = a method to snapshot copies and deltas of volumes between seperate clusters to create backups. Remote copy is ALWAYS asynchronous, therefore does not have latency restrictions. Can be done over LAN or WAN. Can be performed manually or automated via scripts or scheduled recurrance. Remote copies can be configured to be retained based on number of snapshots, or length of time. Snapshots are stored on a SAN/iQ cluster. The snapshot (remote copy) volume can be configured to use Network RAID to protect from node failure of the local cluster that it is stored on.

[Atlantic224]

All

I am very new to this forum and to the P4500. Our setup is a pair of P4500's, one is primary one is setup at a DR site. We are trying to move 2TB of data using remote copy from our DR site to the primary site and it is painfully slow. About 14GB/hour. Any ideas.

Thanks