Tuesday, March 23, 2021

Introducing BGP routing and IPv6 with minimal downtime

Hello guys,

I started working for a company whose network was designed by devs and server admins. I'm spending my days mapping stuff, and 3/4 of what I see makes absolutely no sense. Luckily it's not the production network but a section specifically for R&D where new products/builds are created and tested.

The main issue I faced was that the company had a presence in many datacentres worldwide. Still, only a few sites had the cabinets gathered in a cage, so my predecessor thought of having a 10G link for each rack and put a layer 3 switch with its management interface on vlan1 and assign a public IP. Luckily for me, he was sensible enough to restrict access only from a handful of IPs

I managed to create a consolidation plan and migrating all the racks in a cage; I'm putting a couple of BGP-capable routers in each cage and keep the current switches to handle the L2 comms. On top of that, our main ISP doesn't BGP peering with Google on IPv6, so I'm bringing another one in and configure the IPv6 peering on that interface.

I'm not really on top of my game with BGP peering, and this is the first time I need to to this involving IPv6. Any one of you has had any similar experience? What would you do? In case you already faced a similar challenge, what went wrong, and how did you cope with that?



No comments:

Post a Comment