I'm pretty sure I'm going to at least make someone cringe and give me warnings.... But here goes.
I have 4 machines all running FreeNAS for a small video production company.
On-site, 1 workdisk, 1 archive and 1 backup. I'm working on a new offsite backup which I'd like to get some insight on what's going to be a terrible zfs setup.
Onsite workdisk is a 6 disk Z1 pool with two SSDs for ZIL and L2ARC and 10Gbe, it works very nicely so far for up to 6 editors working simultaneously with 4K video. (At least far better than the Win2k12 storage server that it replaced)
Onsite archive and backup are super micro chassis capable of taking 36 drives.
Onsite archive is configured optimally according to the calculator. 30 X 6TB drives making a pool of 3 Z2 vdevs. Video editors read data out of this machine infrequently while over the course of a year, 6-8TB gets written to it from the workdisk machine when projects are complete.
Onsite backup is non optimal with 24 X 6TB drives in two 12 drive Z2 vdevs. It receives data only from the archive server.
An off site backup that is going into a colocation facility is being built with a chassis that has a 24 physical drive limit. I'm tempted to put a single big fat z2 vdev there just to give us as much disk space as possible with a little bit of redundancy. The off site location is an hour's drive away from the office and connected via VPN with a speed of 15-20Mbps leaving the office.
Rsync is what synchronizes archive to the current backup although I'm thinking about doing zfs replication.
Recognizing that resilvering time when something goes wrong will be affected by the vdev sizes, my work disk and archive are optimally setup while the onsite backup is suboptimal and the proposed off site backup is going to be terrible, I am hoping that having the two backups should cover my butt if any one part of the system goes wrong. Is there any reason I should not setup my offsite backup that way?
I have 4 machines all running FreeNAS for a small video production company.
On-site, 1 workdisk, 1 archive and 1 backup. I'm working on a new offsite backup which I'd like to get some insight on what's going to be a terrible zfs setup.
Onsite workdisk is a 6 disk Z1 pool with two SSDs for ZIL and L2ARC and 10Gbe, it works very nicely so far for up to 6 editors working simultaneously with 4K video. (At least far better than the Win2k12 storage server that it replaced)
Onsite archive and backup are super micro chassis capable of taking 36 drives.
Onsite archive is configured optimally according to the calculator. 30 X 6TB drives making a pool of 3 Z2 vdevs. Video editors read data out of this machine infrequently while over the course of a year, 6-8TB gets written to it from the workdisk machine when projects are complete.
Onsite backup is non optimal with 24 X 6TB drives in two 12 drive Z2 vdevs. It receives data only from the archive server.
An off site backup that is going into a colocation facility is being built with a chassis that has a 24 physical drive limit. I'm tempted to put a single big fat z2 vdev there just to give us as much disk space as possible with a little bit of redundancy. The off site location is an hour's drive away from the office and connected via VPN with a speed of 15-20Mbps leaving the office.
Rsync is what synchronizes archive to the current backup although I'm thinking about doing zfs replication.
Recognizing that resilvering time when something goes wrong will be affected by the vdev sizes, my work disk and archive are optimally setup while the onsite backup is suboptimal and the proposed off site backup is going to be terrible, I am hoping that having the two backups should cover my butt if any one part of the system goes wrong. Is there any reason I should not setup my offsite backup that way?
