Herr_Merlin
Patron
- Joined
- Oct 25, 2019
- Messages
- 200
Hi all,
we have a production NAS with the following layout:
Intel 2U Server:
2x Xeon E5-2680 V2
384GB DDR3 ECC reg
2x LSI TriMode SAS 12G HBA
24x 2.5 Hotswap grouped by 6 disk cages. Each cage is connected to each of the LSI TriMode 12G HBAs
2x RMS-200 8G NV PCIe
2x Intel 80GB SATA SSD for OS installed inside the system
6x STEC 800GB SAS SSD
1x Intel P4x000 M.2 NVMe SSD with PLP
18x 1.8TB 10k SAS HDD (those Seagate ST1800MM0018)
1x Intel Dual Port 10G Networking for NFS
Thats the basic layout of the Hardware.
System hosts one 1 zpool with the following layout:
9x 2-way mirror of the 1.8TB HDDs
1x 2-way of theRMS-200 for SLOG
1x single Intel NVMe for L2ARC
3x 800GB SAS as special vdev for meta data
3x 800GB SAS as dedup vdev
If I remove the SLOG I get the "simple" HDD preformance.. I would assume as those HDDs have themselvs SSD Cache they would be much faster. Is there any way to speed this up abit more?
Remembering a Seagate Laptop HDD with SSD Cache. That one was way faster compared to to regular Laptop HDDs of same size and rpm. Well running windows ofc...
System has 2 datasets on the pool. Both are shared via NFS with VMware.
One "small" dataset with dedup on for test VMs e.g. n Times Client OS VMs etc. Current Compression Ration with dedup and zstd-7 sits at 1.87.
Second "big" dataset is used for a few production VMs with bigger storage needs.
Overall usage of the pool sits at 53%.
atime is off
Record Size is set to 64KB
sync to always
tuneables:
vfs.zfs.l2ar.mfuonly 1
vfs.zfs.l2ar.rebuid_enabled 1
vfs.zfs.l2ar_noprefetc 0
TrueNAS CORE 12.0-U6
TLDR:
How to utilize the 32GB Cache on ST1800MM0018 with TrueNAS?
we have a production NAS with the following layout:
Intel 2U Server:
2x Xeon E5-2680 V2
384GB DDR3 ECC reg
2x LSI TriMode SAS 12G HBA
24x 2.5 Hotswap grouped by 6 disk cages. Each cage is connected to each of the LSI TriMode 12G HBAs
2x RMS-200 8G NV PCIe
2x Intel 80GB SATA SSD for OS installed inside the system
6x STEC 800GB SAS SSD
1x Intel P4x000 M.2 NVMe SSD with PLP
18x 1.8TB 10k SAS HDD (those Seagate ST1800MM0018)
1x Intel Dual Port 10G Networking for NFS
Thats the basic layout of the Hardware.
System hosts one 1 zpool with the following layout:
9x 2-way mirror of the 1.8TB HDDs
1x 2-way of theRMS-200 for SLOG
1x single Intel NVMe for L2ARC
3x 800GB SAS as special vdev for meta data
3x 800GB SAS as dedup vdev
If I remove the SLOG I get the "simple" HDD preformance.. I would assume as those HDDs have themselvs SSD Cache they would be much faster. Is there any way to speed this up abit more?
Remembering a Seagate Laptop HDD with SSD Cache. That one was way faster compared to to regular Laptop HDDs of same size and rpm. Well running windows ofc...
System has 2 datasets on the pool. Both are shared via NFS with VMware.
One "small" dataset with dedup on for test VMs e.g. n Times Client OS VMs etc. Current Compression Ration with dedup and zstd-7 sits at 1.87.
Second "big" dataset is used for a few production VMs with bigger storage needs.
Overall usage of the pool sits at 53%.
atime is off
Record Size is set to 64KB
sync to always
tuneables:
vfs.zfs.l2ar.mfuonly 1
vfs.zfs.l2ar.rebuid_enabled 1
vfs.zfs.l2ar_noprefetc 0
TrueNAS CORE 12.0-U6
TLDR:
How to utilize the 32GB Cache on ST1800MM0018 with TrueNAS?
Last edited:
