Synology HA active-passive performance and failover

I have two Synology boxes (DiskStation DS1817+) that are running in Synology HA mode filled with WD Black 4TB drives in RAID10. They are connected to my cluster over iSCSI providing mainly backup storage, file share (in a separate File Server VM) and hosting a couple of DEV virtual machines and test labs.

The problem is that these boxes run in an Active-Passive mode which is not completely great since I get the performance of a single box only (which is already not sufficient on current stage) and failover takes too much time (around 2 minutes) which cause the services running there to die. Is there any way to switch them to active-active so both boxes are used to speed things up?

Another way I am considering is converting them to RAID0 but it’s obviously a bit scary. Furthermore, I am not sure that 4x1GbE networks will not bottleneck the RAID0 performance in this case. Any suggestions are highly appreciated.

Unfortunately, Synology HA feature does not support active-active mode. If you want to speed things up you might do RAID0 but it’s quite dangerous and will completely load your networks (especially on sequential workloads) while processing backups and large files on a file share. There is an option to use some 3rd party software that will treat your Synology boxes like regular SAN or NAS and do the active-active mechanics on top. In order to do this, you will need at least two servers (but since you’ve mentioned a cluster, I assume you have those). Another good thing about this approach is that failover time will be significantly reduced either.

Here is a great example based on similar DiskStation you have that might give you all the answers you need https://www.starwindsoftware.com/blog/synology-diskstation-ds916-turbocharged-with-starwind-virtual-san

Hope this helps.

Decouple your Synology HA setup, RMA one of these units with "Your HA does not work!", and get VMware or similar vSAN to enjoy local disk performance, and reliability. Whatever you'll do you won't squeeze any reasonable IOPS, and transparent failover time out of your current setup.

P.S. You might try to use IBM SVC (now IBM Spectrum Virtualize) or already referenced StarWind to fix broken HA but that's an overkill IMHO.

https://www-03.ibm.com/systems/storage/software/virtualization/svc/