aravindrajbalaj wrote:Backup is completing but it nearly 15 to 16 hours and data size is just 10GB. Is there a way to reduce the time.
Short answer: Put the source on more spindles.
Real answer: You have to find out whether this is just a lots-of-small-files problem or you are being impeded by the IDB going single-threadedly CPU-bound (or, not unlikely, both). On the source server, use the resource monitor (available from the task manager) and check if during the backup, disk saturation is approaching 100%. If that is the case, you cannot read any faster (from the source hardware in question with the data pattern of a file system traversing backup) and the only way to change that would be making the hardware faster[1]. If the source isn't saturated, maybe the IDB cannot cope with the amount of metadata coming in, which may be either saturating the CPU (or rather, the single core the old IDB ran on), or again the I/O (of the storage backend the IDB resides on). If you see this, and only this, an upgrade to 8.x might help, as the all-new IDB introduced in this version should fix exactly these issues. But plan this upgrade wisely, there are new issues in the box. Ideally, read up here on any thread talking about upgrades to 8.x to prepare yourself for what's coming.
[1] To be more exact, there is another way to speed up reading of this data pattern, at least to some amount: disk block location oriented (sorted) reading. It's something I've never found implemented in an enterprise backup software, though. And it's only helping so much. But it does speed up things, I've written a little C program to demonstrate that (on Linux platforms).
HTH,
Andre.