Two C800 (R1) clusters which are both identically in terms of hardware (100G NICs all around, and 12 x 15TB disks)
They are setup in SMas setup over two 40G lines between the two clusters..
We have two x 100G per node towards the VM hosts (connected to two redundant switches), and two x 100G per node for intercluster traffic.
The ESX hosts also have two 100G NICs each..
Never the less while reading and writing I cannot get the speed over 650MB/sec. which seems kinda low to me? Example copy is from the C800 cluster to a C250 (also with 100G NICs) and back again... does not matter which way, the speed does not get much above the 650MB/sec...
To me this seems a bit slow with such a setup? CPU isn't busy, iops are at about 10K/sec. and latency is 250us.. everything looks like it's taking a vacation 😉
Any suggestions? 🙂 What should you normally see on a C800 with 12 x 15TB disks installed?
#Performance question...
true hare
thorn stirrup
so you are doing SMas where EACH SITE - the network is 100G to the clients and the local cluster switches (- intracluster ) and the connection between the cluster (intercluster) is 40G?
If you are writing to the SMas, you will have to "wait" for the ack on the other cluster
Otherwise, you >should< be able to get better than that but you may need more simultaneous clients
If you are using VMware, esxi v8+ have you tried enabling nconnect?
https://knowledge.broadcom.com/external/article/313464/support-for-nconnect-feature-added-in-es.html
https://docs.netapp.com/us-en/netapp-solutions-virtualization/vmware/vmw-vsphere8-nfs-nconnect.html
true hare
If nConnect is NFS only, it doesn't make much sense in this setup as it is ASA with iSCSI...
One thing I noticed was that the "proximity to" in the groups isn't set to anything... the igroups was setup by OTV as we setup the systems... maybe I should look into setting this correctly? But still even with this, we should be able to max out the 40G line which is 4-5 times the speed we see now 😉
