I'm curious how others are working with customers running code that drop the session with valid BGP attributes. Anyone else monitoring the proliferation of routes with attribute 128? I'm not really in favor of the features vendors have provided, such as this to just drop the attribute or routes. http://www.cisco.com/en/US/docs/ios-xml/ios/iproute_bgp/configuration/xe-3s/... - Jared
On (2013-09-25 11:35 -0400), Jared Mauch wrote: Hi,
I'm not really in favor of the features vendors have provided, such as this to just drop the attribute or routes.
I would encourage customers to require in their transit agreements that bgp updates are not mangled by provider. It would help internally if you have customer backing. I think it's overraction to kill useful features because sometime on some platform there has been NLRI parsing bug which caused issues. Once those filters are deployed there will be strong resistance to remove them. -- ++ytti
I certainly agree. There is a very narrow case for filtering 128 as it's a VPN attribute that should not be in the big-I Internet. Jared Mauch
On Sep 26, 2013, at 4:28 PM, Saku Ytti <saku@ytti.fi> wrote:
Once those filters are deployed there will be strong resistance to remove them.
On (2013-09-26 17:02 -0500), Jared Mauch wrote:
I certainly agree. There is a very narrow case for filtering 128 as it's a VPN attribute that should not be in the big-I Internet.
I can't think of application right now, but I'm not convinced there isn't application for 128 over INET. I know RFC strictly speaks about VPN deployment, but I wouldn't be too surprised if someone would have good use-case for tunneling attributes over INET. -- ++ytti
participants (2)
-
Jared Mauch -
Saku Ytti