Can we estimate time used for arcmain data copy/restore using linear scale from sample data? i.e. 5 gb is copied in 5 minutes then 1 tb in 1000 minutes.
If your network speed is consistent and the same as for the 5 G copy, then yes.
If you are not already, then you may want to use multstream job for Higher volume of data so this will be faster.
as always the usual answer "it depends", some other factors in play
Data/objects make up, multi-stream backups work best with large objects if there are a proportinally more tiny objects in a backup set there will be an adverse affect since there is a constant build-up tear-down process for each object this is where using a single-stream backup for things like views & macro databases may perform better but from a backup set intergrity point of view most people would be grouping their data based on mainly on application/business requirements.
Ultimatley the biggest variant would be if you are running the backup/restore against a quiesent system or not, as backup and restore are I/O intensive operations if they have to contend with other processes for a finite resource.
Thanks. My env will be available primarily for arcmain. Also objects is just whole of databases. So parallel arcmain jobs will give more throughput?