Depends on the type of backups you have. Jason has a long stack of incrementals to apply to a much earlier full backup and does not know it. I use nothing but full FS duplication. It takes time but restores are always valid (though they might not be in MSwin OSs, not very concerned nothing valuable there any more).
Usually the latest backup *is* a full backup and the incrementals are what allow you to go back to an arbitrary point in the past. Or in my case the backup system (dirvish/rsync) uses the filesystem hardlink facility so that *all* backups are full, but identical files are hardlinked to the same underlying copy of the data.
No, I hve a short stack of full backups supporting the current full backup, the decrimentals are handy if there is need for a file from last wednesday, or from last month.
the system picks one of the old ones renames it replaces any files that have changed since it was made with a hard link to yesterdays backup or copy of todays live data whichever is apropriate.
the only thing incremental is the way that it is maintained.
Where are you getting the idea that de-dup somehow increases restore time? Have you looked at any enterprise class backup servers? The de-dup is all controller based, the host just gets a full image back if that is what you request.
You definitely haven't looked at current enterprise class disk based backup servers with data de-dup. They are not even remotely like old tape based full/incremental backups.
The biggest problem on that 56 hour rebuild was the backup did not want to restore properly - using Computer Associates Arcstore and HP Ultrium tape backup. Got 3 of the VMs back no problem - one after a lot of fuss, and the last one had to be rebuilt from scratch, and the SQL database re-imported from DB backups -
There are, sadly, about as many definitions of de-dupe as there are solution providers - and what each one considers de-dupe is SO vastly different from the rest that "de-dupe" really has no real meaning today.
We are currently using VEAM - and no, I do not fully understand how the darn thing works. All I know is it replicated from the operational server to the standby server, then backs up from the standby server to a QNAP NAS, and from there to a WD Go-Drive for offsite storage. I call it belt and suspenders (and a spare set of trousers) - but I will NEVER be down for 2 1/2 days again. The 2 servers are at opposite ends of a 200+ foot concrete block building - and we are looking at moving the standby server into an adjacent building (about 160 feet away - connected by several underground conduits)
People also underestimate the real cost of downtime / data loss on home machines. If you do more than surf and play games it starts to have real costs associated with it.
My home data management scheme:
Primary storage - LG NAS box with two 1TB disks in a mirrored configuration. Fault tolerant and supporting storage for all important files accessed by the 7 or 8 PCs here. Backups of the PCs systems disks are also pushed here periodically. Everything is on UPSes of course.
Offline onsite storage - 1TB Seagate USB3 portable drive, encrypted with TruCrypt and kept in a firesafe. This is a full copy of the LG NAS.
Offline offsite storage - 1TB Seagate USB3 portable drive, encrypted with TruCrypt and kept in a safe deposit box at a bank a few miles away. This is a full copy of the LG NAS.
Periodically the offline onsite drive is retrieved from the safe, updated with a fresh encrypted copy of the NAS and then brought to the bank to swap out with the one in the safe deposit box. The one from the box is brought back, updated and put in the safe.
Simple, inexpensive (~$600 in gear), reliable and easy to manage.
That sounds like the even-odd alternating full backup scheme, with one of the four copies erased and rewritten each time. I do something similar with three drives and data that isn't worth encrypting. If a drive fails while rewriting the erased one I still have one backup. I run it on a laptop which considerably reduces the size of the UPS needed. jsw
Doubt whatever you want. I made the backups, daily incrementals, weekly full backups, monthly full backups sent off site. I examined the data on the backups carefully and redid any backups that failed; about one in twenty. The incrementals were a fraction of the size of the full backups, and included far fewer files. Full backups took a few hours, the incremental backups took about twenty minutes. This was a while back but the fundamental properties have not changed.
Really? Do a bare metal restore and see if that it really true. It is the only way you will really find out. A backup system that is not tested (regularly) is a bad joke. Don't let the joke be on you.
And i'll give you the same advice i just gave Jason. Test it, do some bare metal restores. If that works the way it is "supposed to" you may have support for your smug assurance. Don't be surprised if it doesn't.
You made Jasen's backups? I was not talking about *your* process; I believe you. In fact it seems Jasen uses a scheme similar to mine below (using hardlinks).
Perhaps I should rephrase. Of course it is the changes that are calculated (and transmitted if offsite). But a system is possible where the changes are applied to so as to maintain a "full backup" copy and
*also* recorded in such a way that previous backups can be reconstructed as required. So the most recent backup can be restored quickly and reliably, the usual case. It is the older backups that rely on applying a series of differentials in sequence.
But there are *many* other schemes possible.
Well, one thing that may have changed is an increase of use of hard disks for backup storage. This does enable the use of more efficient backup strategies compared to tape. Such as hardlinks, data deduplication, filesystem snapshots, to name three major ones. Another change might be the ability to efficiently transmit differentials off-site, allowing off-site copies to be maintained automatically.
Every six months a vackup goes off-site and is "installed" on another server. This is the VM backup. All we really nead for a "bare metal". The database backup also gets checked every 6 months or so. A WHOLE LOT easier to test, and less likely to fail, than the old tape system. (which we backed up on DL DVD as well)
ElectronDepot website is not affiliated with any of the manufacturers or service providers discussed here.
All logos and trade names are the property of their respective owners.