General_Mayhem
Cadet
- Joined
- Jul 17, 2023
- Messages
- 1
Hi all, new to truenas in general. Looking for a little advice and information. I have searched and either I lack the context to make use of the information in the KB or it's not there. I appreciate the time taken in any reply.
The use case is primarily file storage, smb shares (10 users in development and LAB environment), a working tier (SSD) and archive tier (HDD), both apps in LAB and development workstations will use the archive tier for incremental backups. Operating degraded while volume rebuilds is acceptable for short periods of time (48 hours)
A trial use case is shared storage for Proxmox cluster, this is not the primary, but if it could work out for our lightly loaded development lab then it would be awesome. (20-50 vms), only 4 of them are io stand outs, Industrial Automation Historians) 65k iops reads and 15k iops writes (both of these are peaks) I assume that this would be on the ssd pool only.
Our existing DAS pool would stay in place until this solution was validated, so this is a pilot not prod.
I have 3 1u servers available for the cluster pilot with each having the following components..
12 core cpu
128gb ram
4x Enterprise 18tb HDD
3x Enterprise 7.68tb SSD
4x Optane 118gb Nvme (typically we reserve 2 for mirrored boot)
2x Consumer 2tb Nvme (not sure if these will stay, optional to replace with optane if design dictates)
2x 10gbe
2x 1gbe
What is most likely to be best for this uses case:
Shared Storage Cluster (does this scale to 3 nodes?)
or
Erasure-coded clusters
With shared storage clusters does scale create the local vdevs in a mirror or are the nodes mirrored?
Will the 10gbe be enough? I would loose the 4x nvme riser card if I had to swap out to 25 or 40.
Kind regards;
The use case is primarily file storage, smb shares (10 users in development and LAB environment), a working tier (SSD) and archive tier (HDD), both apps in LAB and development workstations will use the archive tier for incremental backups. Operating degraded while volume rebuilds is acceptable for short periods of time (48 hours)
A trial use case is shared storage for Proxmox cluster, this is not the primary, but if it could work out for our lightly loaded development lab then it would be awesome. (20-50 vms), only 4 of them are io stand outs, Industrial Automation Historians) 65k iops reads and 15k iops writes (both of these are peaks) I assume that this would be on the ssd pool only.
Our existing DAS pool would stay in place until this solution was validated, so this is a pilot not prod.
I have 3 1u servers available for the cluster pilot with each having the following components..
12 core cpu
128gb ram
4x Enterprise 18tb HDD
3x Enterprise 7.68tb SSD
4x Optane 118gb Nvme (typically we reserve 2 for mirrored boot)
2x Consumer 2tb Nvme (not sure if these will stay, optional to replace with optane if design dictates)
2x 10gbe
2x 1gbe
What is most likely to be best for this uses case:
Shared Storage Cluster (does this scale to 3 nodes?)
or
Erasure-coded clusters
With shared storage clusters does scale create the local vdevs in a mirror or are the nodes mirrored?
Will the 10gbe be enough? I would loose the 4x nvme riser card if I had to swap out to 25 or 40.
Kind regards;
