iSCSI target has limited IO operations queue. Therefore in some scenarios having 1 volume per target instead of 10 can give more performance. In HA cluster during the failover all configuration of one node has to be exported and imported on other server, so switching time depends on number of volumes (switching volume replication) and targets (exporting configuration), so its also a good idea to keep it in reasonable amounts suitable for your hardware. Best to test switching time with some active IO test before going into production.