Here is a problem we are having with media agents at one location. We are seeing backups and restores with wildly different performance for any gateway installed on a Nutanix cluster. An adjacent older VMware cluster has no problems. The two clusters share the same 1G Internet pipe at a colo.
We have another Nutanix cluster at a different location that has pretty decent performance consistently. We cannot find any one thing that might explain the behavior, and the poor and inconsistent performance makes managing the platform very difficult at that site.
We would like to understand why the gateway running on the older cluster (HPE servers and EMC storage) which runs on 1G interconnects, is orders of magnitude faster than some of the gateways running the new Nutanix cluster (HPE DX-380’s) with 10G interconnects.
Running ordinary uploads and downloads on any of the Nutanix-hosted gateways is fast, as expected. So, it doesn’t seem to be Internet, or hardware related. It smells as though Metallic software is misbehaving, but only on some gateways on that cluster.
We have been escalating with the Metallic support but so far there have been no revelations! Our next step may be to install gateways on physical hardware and see what happens, taking the virtual environment out of the picture as far as possible.
Has anyone else worked a problem like this with their environment?
