For me the one to beat is IBM’s backup utility (known for years as ADSM (adstar storage manager) and later called TSM (tivoli storage manager). I’ve never seen a commercial or open source program that comes close.
I’ve also never seen successful backups anyplace that did not have TSM. Usually the backups are corrupt, or nobody knows how to do a restore, or you need 1000x the storage capacity in order to restore the initial backup and all of the incremental backups until you reach a specific point in time.
At places I worked with TSM it was so simple that individual users could fire up a gui and pull files out of the backup pool.
On the backend we had massive IBM tape libraries and it was hypnotic to watch the robot jet around moving tapes in and out of drives and the storage slots. It never stopped moving either, when backups are restored were not happening it was busy consolidating the tapes, making copies of data from tapes that had been used too many times, or preparing copies to be sent off site. It was a full time job for someone to load new blanks when TSM requested and remove the offsite tapes and put them into a box for fedex to pick up. (The one thing that has not changed is that it’s still quicker to send massive amounts of data by fedex then it is to send it over the public internet)
I have used TSM (or ADSM or Spectrum Protect or whatever IBM calls it this week) quite a bit. The basic functionality and performance are not too bad. However, it clearly shows that the software originates in the 1980s. The client is written in C++ and really likes to leak memory. This becomes problematic when backing up more than a few million files. The official "fix" suggested by IBM is to configure a cronjob that restarts the scheduler once a day (seriously).
TSM also has no support for deduplication, so good luck backing up large variable binary files such as VM images or project files (video, CAD, etc).
I’m pretty sure it did originate in the 80s, it had an earlier name than ADSM, then was rebranded back when IBM was going to split itself into “baby blues”, then Lou Gerstner took over and stopped the split-up. Despite its faults it’s still the best I’ve ever encountered.
I’ve also never seen successful backups anyplace that did not have TSM. Usually the backups are corrupt, or nobody knows how to do a restore, or you need 1000x the storage capacity in order to restore the initial backup and all of the incremental backups until you reach a specific point in time.
At places I worked with TSM it was so simple that individual users could fire up a gui and pull files out of the backup pool.
On the backend we had massive IBM tape libraries and it was hypnotic to watch the robot jet around moving tapes in and out of drives and the storage slots. It never stopped moving either, when backups are restored were not happening it was busy consolidating the tapes, making copies of data from tapes that had been used too many times, or preparing copies to be sent off site. It was a full time job for someone to load new blanks when TSM requested and remove the offsite tapes and put them into a box for fedex to pick up. (The one thing that has not changed is that it’s still quicker to send massive amounts of data by fedex then it is to send it over the public internet)