Page 1 of 1

How is data laid out?

Posted: Wed Jun 09, 2010 2:18 pm
by alazanowski
I sent in a support request but i havent heard anything back on this, so im going to check it in with the forums.

Currently we are testing out a procedure where we backup a system using the r-drive image software with no compression set. We then backup the .arc files using netbackup puredisk which segments files into 128kb segments, compares them to what is currently stored, and then only stores unique 128kb segments (deduplication). While I assume there wouldn't be perfect deduplication on these, i am very surprised that there is a 0% deduplication rate on a full backup taken at 12pm, and another full backup taken at 1pm from the same system. Why would the backups appear completely different from each other if no data has really changed on the system in 128kb segments? Even if i have 3-4 of these full backups from the same system, nothing appears similar between them.

Im trying to validate this as a concept to buy a heck of alot of licenses of the product, so please let me know when you can.

Re: How is data laid out?

Posted: Thu Jun 10, 2010 4:09 pm
by Alt
Generally, if nothing has been changed on the disk, the images should be the same, except some small areas in the beginning and the end of the file.
R-Drive Image uses the same technique and splits the data into blocks of app.256kB. If the Backup the useful information only option turned on and something has changed at the beginning of the disk (some cluster might turn from used to not used, or vise verse), and all data may shift, making the image look different for "deduplication".

Re: How is data laid out?

Posted: Wed Jun 16, 2010 12:08 pm
by alazanowski
So is the appropriate size to slice with puredisk to be 256kb? and is that kilobits or kilobytes?

We'll run some additional tests to see this. If a system performs a defrag, will that change the data significantly?

Re: How is data laid out?

Posted: Thu Jun 17, 2010 8:31 am
by Alt
alazanowski wrote:So is the appropriate size to slice with puredisk to be 256kb? and is that kilobits or kilobytes?
It's worth trying. kilobytes.
alazanowski wrote:If a system performs a defrag, will that change the data significantly?
That depends on how the data was fragmented. If seriously, the changes are serious.