« Deduplication and Replication | Main | CDP vs. Backup »

September 11, 2008

Comments

Feed You can follow this conversation by subscribing to the comment feed for this post.

Jered Floyd

Scott,

Owwww. That was painful! I'd complain more, but I'm just as guilty: http://www.permabit.com/videos/video-library.asp

I blame Marketing, they forced me to do it!

Seriously, though, tape's days are numbered. I wish the video explained a bit more about why dedupe to tape is a terrible mistake, but tape can no longer claim a cost advantage. The only thing it's good for is moving a lot of data from point A to point B quickly, and network interconnect is catching up....

Jered Floyd
CTO, Permabit Technology Corp.

Scott Waterhouse

Jared;

Simple reason why dedup to tape doesn't work: tape like streaming. Tape likes lots of big chunks of data sent at high speed.

Dedup is all about lots (LOTS!) of little bits of data. In a 10 TB dedup box, there might be 500,000,000 segments if I did my math right. A good way to get an LTO4 drive to run at 0.5 MB/s is to try to write a bunch of small objects of this size to it without any aggregation by a backup application.

The flip side of this coin is there are few (none? somebody more knowledgeable than me is welcome to chime in here) operating systems that can quickly (50 MB/s +) read and process 500,000,000 small objects.

Net result: dedup has to be done to disk.

The comments to this entry are closed.

Search The Backup Blog

  • Search

    WWW
    thebackupblog