On the incredibly expensive side, you have things like NetApp/etc, and on the free-if-you-can-support-it side is ZFS, which is what I mainly use.īest practice would still be to have a second form of backup, but that second backup can run for longer periods of time in most cases since it's not your primary backup. The only really excellent solution I've ever found at that scale has been snapshot-based replication and manage snapshot retention at both ends. "NAS" style approaches (file based) can start to really fall apart at that scale. I've had to deal with this problem in several cases over the years with absurd numbers of files on a system.
100+TB and tons of files? The problem with huge (tens+/hundreds+ of millions) numbers of files is the metadata hit, rather than the sheer amount of bytes actually transferred.
0 kommentar(er)
