in RAC each instance has a separate redo thread, that has it's own sequence number. Yes your backupsoftware IS backing up redologs from both nodes and this is required for RAC. You don't have duplicate archivelogs because of this, each of those files is unique and must be in the backup. If you connect to node 2 your session is generating redo on instance 2, instance 1 doesn't know of this session in its redo logs. If you would e.g. omit the redo logs from instance 2 from your backup, you could not recover the changes done from this (and all other) session on node 2 in case of a crash. The redo logs from node 1 could simply not know of the changes done.