Unsolved
This post is more than 5 years old
2 Intern
•
326 Posts
0
753
Dedupe
Hi folks
From the deduplication theory, the first or baseline backup would be the longest or real backup time. Subsequently the backup should be faster.
Example (NDMP data)
FULL 1TB of data.. 10 hours to tape
1st FULL of data ... 10 hours to dedupe
subsequent FULL should be less than 10 hours....assuming changes are just 10% and since it's byte level dedupe...backup should take minutes rather than hours or close to 10 hours to complete..
Is this understanding correct? If yes, and if the backup to dedupe still taking same 10 hours, what could be possibly wrong?
Thanks in advance for your sharing...
yagernc
5 Posts
0
February 10th, 2012 15:00
In theory yes it should take less time. Keep in mind it's not simply comparring or looking for previous backups of the same data, but looking for similar blocks already stored on that device.
If it still takes the same time, you could be sending data that doesn't dedupe well. Like compressed data or images/videos etc.. Or you could have other issues like network or system performance etc...
Thierry101
2 Intern
2 Intern
•
326 Posts
0
February 12th, 2012 20:00
Hi
Thanks for reply...
It's a 3 way backup...
Celerra to storage node (DD Boost) and to DD...checking and comparing the blocks should be fast as claimed by EMC on marketing slides using the hash algorithm and pointers without sending the real data over the LAN to compare...we do save on storage space but time is the question mark...