rafalefighter
Cadet
- Joined
- Nov 5, 2022
- Messages
- 8
Hi All,
I'm sorry if similar questions asked before. I'm new to forums. We were using Truenas-core as a backup storage for few years with Dell hardware. Our setup was stable and never had any issues.
But recently we got requirement to find storage solution for our VM storage as we are decommissioning our small ceph storage (6nodes) cluster.
I was checking multiple storage solutions including synology / Dell / Hitachi VSP (we using one in another site) / Hpe Nimble. but the big issue is we cannot get any of these deployed quickly and due to the cost we need approval and things. So our plan is to reuse the existing ceph hardware and leverage Truenas core for VM storage. And later go with enterprise grade storage server like Truenas Enterprise appliances.
We don't have much experience with Truenas as primary VM storage. So we quickly put together single node Truenas-core 12 U8.1 box for testing this.
After the base installation we create LACP with 2x Juniper qfx51xx switches and tested the networking part. all seems working good. Then create one large pool with Raidz2 and export the NFSv3 share to our Proxmox-ve cluster with the NFS number of servers value set to 12. Not added any tunable yet.
Then we move clones of some of the high IO vms to this storage on proxmox (12vms). Everything seems working fine. we did some test like backups and live migration tasks as well. Capacity usage is even less than 2% at the moment due to Lz4. I did scrub task and Smart tests while the VMs are running and seems no much Impact from that as well.
Now I have few questions
Thanks
I'm sorry if similar questions asked before. I'm new to forums. We were using Truenas-core as a backup storage for few years with Dell hardware. Our setup was stable and never had any issues.
But recently we got requirement to find storage solution for our VM storage as we are decommissioning our small ceph storage (6nodes) cluster.
I was checking multiple storage solutions including synology / Dell / Hitachi VSP (we using one in another site) / Hpe Nimble. but the big issue is we cannot get any of these deployed quickly and due to the cost we need approval and things. So our plan is to reuse the existing ceph hardware and leverage Truenas core for VM storage. And later go with enterprise grade storage server like Truenas Enterprise appliances.
We don't have much experience with Truenas as primary VM storage. So we quickly put together single node Truenas-core 12 U8.1 box for testing this.
- 2x Intel(R) Xeon(R) CPU E5-2680 v4 @ 2.40GHz (2 Sockets) - 28 cores 56 Threads
- 128GB ECC RAM (HPE)
- 6x SAS 3.2TB SSD Drives (HPE)
- 2x HPE Ethernet 10Gb 2-port 560SFP+ Adapter
- 2x PSU
After the base installation we create LACP with 2x Juniper qfx51xx switches and tested the networking part. all seems working good. Then create one large pool with Raidz2 and export the NFSv3 share to our Proxmox-ve cluster with the NFS number of servers value set to 12. Not added any tunable yet.
Then we move clones of some of the high IO vms to this storage on proxmox (12vms). Everything seems working fine. we did some test like backups and live migration tasks as well. Capacity usage is even less than 2% at the moment due to Lz4. I did scrub task and Smart tests while the VMs are running and seems no much Impact from that as well.
Now I have few questions
- Is this hardware combination good choice for Truenas-core 12 ?
- ISCSI or NFS ?
- Mirrored vdevs or raidz2 pool ?
- How will be the recovery process if server crash/reboot ?
- How bad the VM iops if storage is more than 50% full and scrubbing ?
- Any checklist that I need to follow before deploy Truenas as VM storage ?
- Any risks if we go with this setup ?
Thanks