Home
| Calendar
| Mail Lists
| List Archives
| Desktop SIG
| Hardware Hacking SIG
Wiki | Flickr | PicasaWeb | Video | Maps & Directions | Installfests | Keysignings Linux Cafe | Meeting Notes | Linux Links | Bling | About BLU |
On 12/12/2011 11:06 AM, Mark Woodward wrote: > In your example, a duplicate reducing backup would ignore most of the changes. > > > Edward Ned Harvey <blu at nedharvey.com> wrote: > >>> From: markw at mohawksoft.com [mailto:markw at mohawksoft.com] >>> Sent: Sunday, December 11, 2011 2:48 PM >>> >>> I will argue that an rsync will NEVER be more effective unless you >>> actively wipe the blocks where a file once existed. >> for (( i=0 ; i<200 ; i++ )) ; do >> mkdir temp >> cp datafile temp >> run_test $i >> testresults.txt >> rm -rf temp >> done >> >> In this case, rsync is what you want, because it ignores files that don't >> exist. But a block level backup will backup all the blocks that were ever >> contained in any of the (now removed) copies of the datafile. >> >> I don't know what users you support, but I support engineers who run this >> type of test all the time. They create test work dirs, they perform >> volatile work in there, store the results of the test, and remove their >> scratch dir. >> >> The block level backup you're talking about is great, under the assumption >> that you basically just add data to a filesystem. It's terrible when you >> add & remove data from the filesystem. I stand by my claim: Important to >> know if it's suitable for your purposes, whoever you are, the consumer who >> might consider using this. >> I have always preferred a file-oriented backup approach, but I have also been burned. I used to build tarballs until my backup of my home directory placed a VM in the tarball on a 32-bit Linux, and the drive where my home drive was crashed. I was able to restore everything up to the VM that was larger than 3GB. Eventually, I paid to extract the data from the hard drive because the I lost my email archives and checkbook. With today's larger HDs and/or inexpensive NAS systems, like the WD MyBook, you can use rsync's --link-dest so you can have the equivalent of both a full backup and an incremental backup. At work, I use rsnapshot to back up our systems and keep about 20 days online, but New York backs us up daily and performs a tape backup so in the case of a disaster that would knock out our systems, we would lose minimal data. The upside of this scheme is that the most frequent need for a backup is that someone accidentally deletes a file. At home, I do a similar thing except I don't do a remote backup (I know very well I should :-). The disadvantage of a file-oriented backup is if you have to do a full system restore, but this is rather rare. At home, if I were to lose my primary HDs I would essentially have to spend time to rebuild the entire system. -- Jerry Feldman <gaf at blu.org> Boston Linux and Unix PGP key id:3BC1EB90 PGP Key fingerprint: 49E2 C52A FC5A A31F 8D66 C0AF 7CEA 30FC 3BC1 EB90
BLU is a member of BostonUserGroups | |
We also thank MIT for the use of their facilities. |