Vlan based VMs, which device is configured as the default gteway? If it is a L3 switch or FW, then static routing to ESG router for the Vxlan port groups may be needed. Is DLR used for the topology?
In VLAN based VMs gateway is configured core L3 Switch of DC – B, No firewall
in between both environment. NSX infra VMs gateway is ESG devise IP (DC - A) and
DLR also available in my infra we used
DLR to communicate different subnets.
Where we need static routing for this ?
If DC-A and DC-B are on the same location, there could be a common vlan between the ESG External Interface and L3 Switch(Router). This vlan (Blue subnet ) is generally named ESG-External subnet, and the Classical Vlan Port Groups and vXlan based logical switch portgroups need to communicate passing through L3 Switch(Router) --> ESG --> DLR, and the reverse path DLR-->ESG-->L3 Switch. The DC-B Vlan Based Port groups could be considered as an External Network, although the Vlan based VMs and Vxlan based VMs could be on the same Physical ESX host. Vlan based port groups should first reach their default gateway L3 switch.
For this to achieve the routing table of the L3 switch should include an entry for the Logical Switch Vxlan Subnets pointing to the ESG External Interface as the next-hop. This could be OSPF and BGP, but for the simplest topology as below diagram static route is sufficient. This static entry should make the L3 switch forward the Vxlan port group directed packets towards the ESG.
ESG is similar to the L3 switch or router, it has its own routing table. ESG routing table needs at least 2 entries, 1 default route towards the L3 switch, and similar to L3 switch, static routes pointing south towards DLR External IP Address which is in the generally another Logical switch (Black VXLAN). These static entries would make ESG forward the Packets towards DLR.
DLR Receiving the Packets knows how to send the packets to Vxlan VMs because they are already directly connected. DLR Does not need static routes for the VMs, but it needs a default route pointing north to the ESG Internal IP address (Black Interface of ESG) as the next hop.
If the topology is more complex (i.e if there is an mpls cloud between DC-A and DC-B, or ECMP Load balancing is used for ESG HA Redundancy), then static routing would not be sufficient. Bridging could be helpful between a Vlan-Based Port group VMs and Vxlan-Based Logical switches, if both VMs share a common IP Subnet or for P-V conversion.
More detailed could be as below links: