Hub-and-spoke routing over Metro-E handoff

[will.r.french]

Good day everyone,

I have inherited a network that is configured, how shall I say, sub-optimally. I will describe the current landscape, and I'm seeking suggestions as to a plan of attack. Hopefully during this process I'll be able to grow in my knowledge and experience as I implement whatever is deemed the most appropriate solution.

At this time, I have 3 branch offices that connect to my central site over metro-ethernet. My central site hosts my data-center to which of the sites use for server purposes (domain controller, roaming profiles, SAN, etc). Currently, we have an end-to-end VLAN configuration. Obviously my first goal is going to be segregating the 4 sites into a layer 3 routed configuration, as spanning tree doesn't care for traversing the WAN, and it'll make the network as a whole more efficient (i can only assume).

My first (and main) concern is how the metro-e will manage the flow of packets from my core switch to the spoke core switches of my MAN. Perhaps I'm overthinking this, but I'm curious about how it works (or would work). Currently, I have a single port on my core switch that all traffic for all 3 of my spoke sites flows through. This makes it so that when I do a CDP neighbor, I have three devices (the core switch for each spoke site) appearing on one port. If I were to convert this to a layer 3 link, with each spoke site using it's own IP address space, would that mean that all traffic routed to one site will also be received at the other two, and if so, would those two other spoke sites suffer performance wise from receiving and dropping these packets?

Currently everything (and I mean EVERYTHING) is routed through my core router (i.e. all the SVIs are configured within my data center, so literally all of the traffic from one VLAN to another passes through it), so I essentially don't have a core routing protocol functioning. I'm considering OSPF as it seems to be the most versatile, though with only three spoke sites, versatility isn't exactly in high demand. It'll be a relatively simple configuration for a routing protocol, as far as that goes.

Can anyone clarify my understanding on how a routing protocol will handle the above configuration concepts?

Thanks

W.F

[Vishesh Verma · Accepted solution]

Hi Will,

                              +------------------+             +----+

                              |                  |             |    |

                              |                  +-------------+ B1 |

                              |                  |             |    |

                              |                  |             +----+

                              |                  |

    +----------+              |                  |

    |          |              |                  |             +----+

    |          |              |                  |             |    |

    |    HQ    +--------------+      Metro-E     +-------------+ B2 |

    |          |              |       Cloud      |             |    |

    |          |              |                  |             +----+

    +----------+              |                  |

                              |                  |

                              |                  |             +----+

                              |                  |             |    |

                              |                  +-------------+ B3 |

                              |                  |             |    |

                              +------------------+             +----+

Think of Metro-E cloud as a switch.

Right now you must have these interfaces configured as trunks and from each branch you must be carrying Layer2 traffic up to the HQ and HQ router being a multilayer device doing the routing/swtching back to other branches. Metro-E cloud is right now switch trunking all vlans

If you choose to change these interfaces to routed/Layer3 interfaces all of them should be in the same subnet, the Metro-E cloud would act as a switch with all interfaces in a single VLAN and all devices HQ, B1, B2, & B3 would be neighbors to each other and simply share the routes.

So if traffic comes from B3 location with a destination of B1 it would choose B1 as the next-hop, based on the routing protocol and would forward the traffic directly to it.

-Vishesh

[will.r.french]

That is exactly what I was hoping, and now that you explain it, it makes perfect sense. Thank you for explaining it so concisely.