The ZFS filesystem can use a log device (SLOG/ZIL) for where write for sync based i/o filesystem metadata can be mirrored from the system memory to protect against system component or power failure.
It is for Sync LOG based writes and acts more as a journal device, although it can help accelerate small transaction based I/O like those in database workloads.
NOTE: The ZIL SLOG device is not a write cache that will buffer all I/O's written to the Quantastor Storage Pool.
In most cases, a pair of SSD's per storage pool is enough for the ZFS Intent Log (ZIL). The ZIL is actually not a write cache but an intent log (hence the name).
The ZIL allows writes to go to the SSD's first, which allows more efficient write speeds. When writes come in from clients, the intended writes are stored in RAM and the ZIL simultaneously and then flushed to spinning disks over time from RAM. Once all intended writes are stored on the ZIL devices, a confirmation is sent back to client(s) that the writes have succeeded.
Since the configuration of ZIL is usually a mirrored pair of SSD's per pool and need to be SAS devices
This is because the ZIL is meant to be stable storage that will not be lost in the event of a system failure such as a power outage. Upon a power outage event, for instance, the ZIL would be "re-played" to write all confirmed writes to the spinning disks before making the pool available for usage once more.
If this is an HA environment we need to be able to access the ZIL devices from both head nodes and can only do that with shared storage.
More detail on the ZIL SLOG is available here:
We do support SSD cache tiering for writes and reads via the LSI and Adaptec controller hardware SSD caching options "CacheCade/MaxCache".
These options are recommended if you are looking for the ability to burst IOPS/I/O to the storage without having to always go to the backend platter disk.
More details on this is available at the below links: