Hello,
I have a server setup in an academic environment, 9 nodes, ceph (40gbit network), 1TB ram in each, 2 x socket (64 cores) in each. One of the servers has been set up with sandboxed networks for students, each with an "offline" vmbr to practice classes with. However, we have a lot - 100 vmbrs defined on a single server, serving ~150 vms.
We're experiencing slowdowns in the VMs (all linux based), jerky mouse movement, slow typing etc. But the server dashboard is showing the cpu/ram/network throughput as fairly low. Could the large number of bridges cause such a slowdown, and if so, would vlans on the server be more lightweight on resources?
Appreciate and help or tips folks have. Thanks!
I have a server setup in an academic environment, 9 nodes, ceph (40gbit network), 1TB ram in each, 2 x socket (64 cores) in each. One of the servers has been set up with sandboxed networks for students, each with an "offline" vmbr to practice classes with. However, we have a lot - 100 vmbrs defined on a single server, serving ~150 vms.
We're experiencing slowdowns in the VMs (all linux based), jerky mouse movement, slow typing etc. But the server dashboard is showing the cpu/ram/network throughput as fairly low. Could the large number of bridges cause such a slowdown, and if so, would vlans on the server be more lightweight on resources?
Appreciate and help or tips folks have. Thanks!