Hi all!
First off, my current hardware is as follows:
5x3TB 7200 RPM drives (RAIDz1)
16GB RAM
Pentium G2130
Originally I was using my NAS primarily for general file and media storage, but now have a few ZVOLs mapped to iSCSI LUNs for VMware and for use on a few workstations. I've found that while sequential performance can be quite good, fragmentation is beginning to be a bit of an issue and IOPS are quite horrible, especially for virtual machines. In addition, the redundancy of RAIDz1 is poor.
For all these reasons, I would like to convert this to mirroved vdevs. I will add a 6th identical HDD and create 3x mirrored vdevs.
I realize I will need to destroy the vdev entirely and recreate the new ones from scratch. My question focuses primarily on what the best way to back up all the zvols in the dataset. I have 6 zvols mapped to iSCSI LUNs, 1 jail for Plex and one dataset for media files. Out of all this, I have about 2.2TB of actual data.
One possibility would be to create a snapshot and then replicate the datasets/zvols. I don't have a spare ~12TB of storage though. Is it still possible to replicate to, say, a 4TB storage device with all these zvols? Will it be able to understand that most of this is empty space and do over-provisioning? If not, what other options are there other than building another identical NAS for this purpose? Again, I only actually have 2.2TB of data.
Also, before anybody asks, I do NOT currently back up all the data which is why I need to find a way to back it up now. I only back up the important stuff and leave the less important stuff to redundancy/snapshots.
Any suggestions are warmly welcomed :)
First off, my current hardware is as follows:
5x3TB 7200 RPM drives (RAIDz1)
16GB RAM
Pentium G2130
Originally I was using my NAS primarily for general file and media storage, but now have a few ZVOLs mapped to iSCSI LUNs for VMware and for use on a few workstations. I've found that while sequential performance can be quite good, fragmentation is beginning to be a bit of an issue and IOPS are quite horrible, especially for virtual machines. In addition, the redundancy of RAIDz1 is poor.
For all these reasons, I would like to convert this to mirroved vdevs. I will add a 6th identical HDD and create 3x mirrored vdevs.
I realize I will need to destroy the vdev entirely and recreate the new ones from scratch. My question focuses primarily on what the best way to back up all the zvols in the dataset. I have 6 zvols mapped to iSCSI LUNs, 1 jail for Plex and one dataset for media files. Out of all this, I have about 2.2TB of actual data.
One possibility would be to create a snapshot and then replicate the datasets/zvols. I don't have a spare ~12TB of storage though. Is it still possible to replicate to, say, a 4TB storage device with all these zvols? Will it be able to understand that most of this is empty space and do over-provisioning? If not, what other options are there other than building another identical NAS for this purpose? Again, I only actually have 2.2TB of data.
Also, before anybody asks, I do NOT currently back up all the data which is why I need to find a way to back it up now. I only back up the important stuff and leave the less important stuff to redundancy/snapshots.
Any suggestions are warmly welcomed :)