How does teradata ARCMAIN perform a back up.Does backup take place in parallel? What is the difference between All-AMP backup and Cluster Backup.Which is the best amongst the two or when each one of them will be useful?
Cluster backup allows multiple ARCMAIN client processes to be executed in parallel. Each client process writes to its own archive. An all-AMP backup executes in a single client process and produces a single archive. A cluster backup is potentially much faster, but it is more complex to manage the multiple archive datasets.
Backup is parallel on the Teradata side, but each ARCMAIN writes a single backup file on the client. If you have multiple backup devices available, you generally want to run multiple ARCMAINs in parallel. The simplest approach (and the one I recommend people try first) is to divide the tables/databases to be backed up into groups that have roughly equal backup sizes and run multiple all-AMPs backups.
But what if you have a single table that by itself is far larger than the desired backup size for each ARCMAIN? In that case, CLUSTER backup gives you a way to get the client-side parallelism. CLUSTER backup may also be useful if the list of tables to be backed up changes frequently and you have a single database that is much larger than the desired backup size.
If you run a set of CLUSTER backups one at a time sequentially, it will be no faster than running an all-AMPs backup. And don't forget that for CLUSTER backup you'll also need to have a separate DICTIONARY backup.
Can you share some tips or scripts how can we divide the tables in a database to be backed up into groups that have roughly equal backup sizes. For example database size is 10TB and has around 600 tables and we want to divide the backup of the complete database into 5 parallel jobs. Please share some tips or code. We are using Net Backup