I need some help understanding if Linux Bridge(default config) for Proxmox and what its limitations are, specifically how well it is able to handle heavy volume traffic, I am running multiple VMs that carry sip calls.
SIP traffic is split across multiple ports and each channel sends large number of tiny pockets, we are talking tens of thousands every second across multiple vms. Data volume is not large maybe 5 Mbps down and 3up but this traffic is very granular.
I am starting to suspect that having multiple VMs under same bridge is causing performance problems, so naturally I am trying to understand underlying systems involved with routing traffic inside proxmox.
How are resources allocated to routing and are there options to dedicate resources to this process?
On same note using firewall for filtering this type of traffic may also develop performance problems
What tools are available to monitor and get better understanding on what delays maybe happening inside the bridge?
Sip traffic is very time sensitive and even small delays may cause the receiver to simply "move on"
I am just a bit in the dark about how robust basic Linux Bridge is in handling sip traffic in general and especially when multiple VMs drop sip traffic in to same Bridge for routing.
I am exploring Open vSwitch as well
I will test passing pcie network card directly to few vms on the machine to see how and if performance and stability changes.
SIP traffic is split across multiple ports and each channel sends large number of tiny pockets, we are talking tens of thousands every second across multiple vms. Data volume is not large maybe 5 Mbps down and 3up but this traffic is very granular.
I am starting to suspect that having multiple VMs under same bridge is causing performance problems, so naturally I am trying to understand underlying systems involved with routing traffic inside proxmox.
How are resources allocated to routing and are there options to dedicate resources to this process?
On same note using firewall for filtering this type of traffic may also develop performance problems
What tools are available to monitor and get better understanding on what delays maybe happening inside the bridge?
Sip traffic is very time sensitive and even small delays may cause the receiver to simply "move on"
I am just a bit in the dark about how robust basic Linux Bridge is in handling sip traffic in general and especially when multiple VMs drop sip traffic in to same Bridge for routing.
I am exploring Open vSwitch as well
I will test passing pcie network card directly to few vms on the machine to see how and if performance and stability changes.