Storage Virtual Machines (SVMs) and Networking
What this domain tests
SVM and networking items ask you to place LIFs correctly, explain how IPspaces and broadcast domains isolate traffic, and debug a client that cannot reach a mount point.
Core rules
- An SVM (vserver) is a logical server with its own volumes, LIFs, namespace, protocol services, and export or share policies. The cluster itself is an admin SVM named
cluster. - A LIF has a role, a home port, a current port, a failover policy, and a failover group. It can only fail over to a port inside its broadcast domain that belongs to its failover group.
- An IPspace is a separate address space. SVMs in the same IPspace can reach each other with data LIFs; SVMs in different IPspaces need intercluster LIFs and peering.
- ONTAP 9.7 and later support SVM-scoped routing, so each SVM has its own routing table rather than relying on the cluster table.
- Physical building blocks are VLANs, interface groups, and portsets; an LACP interface group requires matching switch configuration.
Worked example
A Linux host times out mounting 10.20.1.15:/vol1. Check network interface show -vserver svm1: the data LIF is admin up but currently on a port in a different VLAN than its home port after a failover. Confirm network route show -vserver svm1 has a default gateway, then check the export policy rule order - the first matching rule wins, and a broad 0.0.0.0/0 deny placed above the host's rule blocks it. Fix the failover target or broadcast domain, not the client.
Traps
Changing an export policy and forgetting that no reload is needed but rule order matters. Creating a data LIF on a port outside the SVM's allowed broadcast domain. Forgetting that an SVM in a different IPspace cannot simply route to a peer SVM.