is there any limitation for the Veritas Storage?

[rashed_wardi]

Hi;

I am new in symantic products;

we have veritas storage foundaaion HA 6.0.1 for linux ; we want to make a cluster consists of 4 servers, each two are located in deferent city, the giographical distance between them is more than 1000 km ; the customer was inquiry wether is any limitation related to distance, note, there is 1 Gb link between them.

sorry if this like stupid question

thanks in advance;

[mikebounds · Accepted solution]

There limitations in respect to distance are the latency:

1. Heartbeats:
   A single VCS cluster uses at least 2 independent links to communicate between nodes and to detect if a node goes down using a Low Latency Protocol (LLT) and large distances mean the latency is too high
    
2. Storage:
   If using shared storage, then latency becomes an issue, but this is more an issue for the application/user using the storage then the cluster.

So for a SINGLE VCS cluster where the 2 nodes are in separate sites, this is called a Campus cluster, and the VCS admin guide says:

You must have a single VCS cluster with at least one node in each of the two

sites, where the sites are separated by a physical distance of no more than 80

kilometers.

So you will need to use the "Global Cluster Option" GCO with replication. 

So you will have 2 nodes in one cluster contected via GCO (using an ICMP link) to the other 2 nodes in a differerent cluster in the different city.  Then you will need to use replication to replicate the storage - the replication can be Veritas Volume Replicator (VVR) which replicates over IP, or one of several other array or database replication products (like SRDF or Oracle Data Guard) that there is a VCS agent for - choose Agent Type of "Replication Agent" at https://sort.symantec.com/agents to see a list of available VCS agents.

Failover detection time for a node going down in your 2 node cluster will be 15 seconds by default (timeout of LLT links) and this failover is usually automatic, but failover detection for a DR situation when you loose a cluster will be a few minutes by default for the timeout of the ICMP GCO links and this falover is usually manually initiated (but can be configured as automatic) as the network links are not usually reliable enough.

Mike

[rashed_wardi]

thank you Mike for your replay;

but we have  already  this license 

SKU	Description	QTY
License
7I03FZF0-ZZZES	VRTS STORAGE FOUNDATION ENTERPRISE HA 6.0 S64LNX 10 SPVU STD LIC EXPRESS BAND S	1

is it enogh to support what did you suggest ?

[mikebounds]

Regardless of whether you have two 2-node clusters or one 4-node cluster you need an SFHA licence for each node.  I don't know much about group licences, so I don't know if the "10" in your licence means you are licenced for 10 nodes.

If you need a VCS replication agent, regardless of whether this is for use within a single cluster  (Replicated Data Cluster - RDC) or between 2 clusters, which you need if you want VCS to peform the replication actions (make DR writable and reverse roles if possible), then you need a DR licence.  You may also need a GCO licence, but I think a DR license and GCO license may be one in the same.

If you want to use VVR then you need a VVR licence, otherwise you need to to purchare a replication product which would probably be the replication product for your array, assuming each site has the same storage array.

Mike