antrea-live: network topologies
Opened this issue · 1 comments
Xavier reached out, please add details here, will be a fun show
So the idea would be to discuss ways to automate cluster network segregation in vSphere environments when using CAPV to deploy workload clusters. Which I suppose also applies to TCE.
Other CAPI providers integrate tightly with the network stack such as AWS, Azure, OpenStack and others in which you get cluster isolation with dedicated VPCs, T1 routers...
Manually adding vdp, segment, t1 and so on for each workload cluster to deploy isn't very cloud native and having all clusters on a flat network isn't ideal.
NCP (NSX Container Plugin) seems to be the closest option to achieving this at the namespace level but the complexity of it and NSX-T being a requirement (obviously) make it a second class choice.
I'd like to explore what options there are to solve this and if any best practices currently exist / what the current state of the art is.