Hi,
We are testing our new scale out file server and are having major issues with performance.
We have 2 server clustered running the sofs attached to a shared jbod with 8x 4tb sas drives. We have pooled and added the disks to a csv and begun hosting test vm's on the new share.
When testing the volumes locally we get 1,000mb/s of throughput read and write in a spanned volume.
When testing accessing the same volume via the sofs share we are seeing 400mb/s of read and 10-50mb/s of write!!!
This performance is completely unacceptable. Is this due to the sofs forcing write through and no write caching?
All servers have quad port 10gb NIC's with SMB direct. The NIC's and network infrastructure have been tested and all run at line speed. The performance dip is clearly as a result of the sofs tech.
If we create a standard share and copy between servers we get full speeds again.
All servers have all the latest updates and hotfixes applied.
Any suggestions as this is going to completely kill and hopes we had of using this technology.