AFF, NVMe, EF-Series, and SolidFire Discussions
we have an EF-280 wit 48 SSD Disks.
We created 2 Volumes - Raid 6 - workload: other workload
"each" shelf one volume
directoly connected to DataCore Server (Windows) via 32 GB FC
These volumes are in an diskpool on Datacore served to ESXi Hosts
We got 150 MB/s and ~ 2000 IO/s
On other existing storages - like QNAP with SSD via ISCSI to Datacore -> ESXi - we got around 20.000 IO/s
So - what is wrong in our configuration ?
It's possible that you are experiencing performance degradation as a result of delayed ACK being enabled on the host-side. This is referenced in the KB available at https://kb.netapp.com/app/answers/answer_view/a_id/1074155/~/performance-degradation-with-data-assurance-enabled-volumes-and-iscsi-. Can you try following the steps to disable delayed ACK and verify if performance has improved?
we are using the FC Adapter directly connected to the Datacore Server.
We do not use ISCSI
So - should we test this ?
I apologize, as I overlooked that statement earlier in the initial post. Please disregard my post about iSCSI.
Has a support case for this issue also been initiated with DataCore?
If you haven't contacted Support (best way forward), you could create a small disk and present it directly to a non-Datacore FC client, and perform IO from it.
Maybe also check clients and FC switches for errorrs or other problems with the ports/cables involved.