Log in to ask questions, share your expertise, or stay connected to content you value. Don’t have a login? Learn how to become a member.
Hi,I'm new with metro E technology and i'm trying to understand and building a simple topology in order to comprehend how metro e works.Let's say i have this simple topology shown belowI want my VMs from Site A can talk and communicate with VMs in Site B via Metro E. i read that metro e works like connecting two switches but only in different sites. So i'm thinking that i can route my vlans from Site A to Site B and vice versa using L3 protocol (Normal Static Route) using P2P between two switches. From my understanding, is that correct how that metro E works? Can i just simply put P2P IP in those 2 switches and route my desired vlans to P2P next-hop of the switch in Site B and vice versa?Or is it only works in L2 Protocol where the interface should be in native vlans and consist of trunk vlans??Please enlight me from this simple example.Thank you.
------------------------------KARANG DIKA KUSUMA------------------------------
Metro ethernet for just two sites can be looked at as a giant ethernet cable that you plug into a port on each location.You can choose to configure the connected port layer 2 or layer 3 as you wish. I've seen customers use both methods depending on their needs.Full layer 2 can be either a trunk port with multiple vlans bridging the two sites or a simple untagged interface bridging a single vlan between the two devices. This is typically used by designs wanting layer 2 adjacency for redundant systems at both locations. A lot of older technologies want servers to be on the same broadcast domain for the cluster configurations to communicate. The disadvantage is that problems in a single broadcast domain/vlan can create problems at both sites at the same time.Layer 3 deployments are typically used to separate the failure domains between the two sites. These will generally use a routing protocol to facilitate communications. OSPF if a general flood and full connectivity is desired as a simple setup. Or the use of BGP if more control and policy would be helpful to apply between the locations.
Hi @spuluka Thanks for your answer.So based on your explanation from my topology, i could use simple static routing between sites to connect different vlans from both sites then? But i have more questions tho- If i decide i want both sites have the same VLANs from each sites, what should i do? Example like, Site A have VLAN 10,20,30 and Site B have VLAN 40,50,60 and my main apps in Site A are hosting on VLAN10, and i want to migrate my main apps to Site B with the same IP configuration (Hosting on VLAN10). How can i put VLAN10 in Site B but also Site B has its own other VLANs gateways that is communicating with Site A? Do i need to use L3 Protocol alongside with native vlan configuration? Or is there any other way to solve that?
When you decide to stretch the same vlan across multiple sites you do need to consider where the gateway for that vlan is and how that will affect traffic for those vlan resources.In your example, when you add vlan 10 to the second site you can move the desired server. But the gateway for vlan 10 is still at the original site so all layer 3 traffic is bouncing to that other site and then onward to the target instead of routing locally.You also need to consider what happens to resources in vlans stretched between two sites when the link, port or hardware switch that connects the sites fails. This will isolate those devices cut off from their gateway at the opposite site.Typically I only see layer 2 stretch used when required by legacy style clustering applications need that to failover between sites. Otherwise the ease of moving the server from site A to B without ip change is out weighed by the potential problems from either vlan isolation or accidental looping failure domains.