When it comes to creating SVM's is it possiable to create a SVM on node 1 of a cluster and allow it to access data on the partner node? Reason that im asking this is because I heard that some end-users where creating 1 SVM and using it cluster wide. ( accessing aggergates from both nodes )
Can this be done? Is there any documentation out there? Is there any benefit to having 1 SVM per cluster verse 1 SVM per node.
Creating an SVM is a cluster-wide operation. You cannot create and SVM on a single node, but you could configure it to only serve data from one node/aggregate or multiple. When you create a SVM, you must choose on which aggregate the root volume of that SVM will reside and by default it will have access to serve data from all aggregates of all nodes in the cluster. If you specify a list of aggregates that this vserver has access to, it removes access to other aggregates. Ex. :: >vserver modify -vserver "SVM_Name" -aggr-list "aggr_name".
Upon creating a SVM you can choose to create data lifs and management lifs (logical interfaces) that are associated to one or more physical interfaces or a group of interfaces. As an example you could create a SVM to serve only data from one aggregate and the data lif to be associated only to interfaces on that specific node. This would serve data from only one node and aggregate. On another note, you could serve data from a lif associated to a different physical port that doesnt belong to the node where the aggregate is. This could happen for failover reasons if the physical ports of the node go down or by mistake if misconfigured.