Jeffrey Haas

@mcr314@todon.nl This is the time of year to remind you that bats behave strangely when infected with rabies.

There's been some recent discussion covering #bgp routing about the ORIGIN attribute. Pointers to a few items for that discussion follow in the replies, but let's lead with the nice Cloudflare post.[1]

One track follows the fact that RFC 4271 says you SHOULD NOT be changed after it's been originated.[2] But, clearly, that's happening.

Another track has been "we should deprecate this!"[3][4] Good luck with that. Much of the headache for BGP is dealing with incremental deployment. This is baked deep enough into the core protocol that any effort to remove it either at the PDU level or the route selection level sufficient to make a difference isn't something I find likely to happen before I retire.

The missing bit of the conversation is why we're seeing this manipulation at all. The simple answer is that when all you have are hammers, every problem resembles a nail.

Service providers have a very small number of tools in their toolbox to light-handedly shape in-AS traffic. The toolbox to change how traffic is attracted to their AS is just as clumsy.

The subset of BGP's route selection tools available is shown in my very Crayola palette.

The #ietf126 agenda for the IDR working group that standardizes BGP has been posted. It's going to be a busy week! #ietf

@sudorandom@infosec.exchange A hazard of living too long is developing cranky opinions about what you dislike about file formats. Discussing CBOR is for a different day.