For the 4th time to this mailer. What do you do with shipping routers as of 10 years back that have Redirect enabled by default because of the SHOULD in RFC 2461 and RFC 4861? Why is this point so hard to understand or being ignored?
Hemant -----Original Message----- From: [email protected] [mailto:[email protected]] On Behalf Of Christopher Morrow Sent: Thursday, August 19, 2010 5:55 PM To: [email protected] Cc: [email protected]; [email protected] Subject: Re: Router redirects in Node Requirements document On Thu, Aug 19, 2010 at 4:22 PM, <[email protected]> wrote: >> > Redirects are a key part of the Internet architecture. Always have >> > been. >> >> Not sure if you actually looked at the configuration sampling I posted, but >> redirects are not actually used in networks these days. The only places >> where i've seen it used are in "hacked together" networks and were removed. >> While perhaps useful in the 80's and early 90's, there have been numerous >> cases where redirects have been harmful to networks that I've operated. > > Agreed, redirects should *not* be enabled by default. > wow, longest thread about 2 words... ever. Please take my vote as: 1) redirects MUST be implemented. I don't like them, they don't have auth info in them, but I can see cases where they may be useful. 2) redirects MUST NOT be on by default I can see a vendor deciding that on platform X they choose to enable redirects as a default. I hope that no router with more than 2 interfaces on it, and meant for 'not in the home' usage would choose this path. editorial-foo: Quite a bit of the conversation seems like particular use cases being abused for a point. I can see that for large/core network devices there is no need, and significant complexity in maintaining codebase with redirects. In these places I'd prefer to just not have it included at all, but I definitely don't need it on by default. For enterprise-edge, CMTS, wireless deployments I can see that using redirects may be more useful (leaving aside the issues Jared brings up about 'well designed' network architectures) so having the codebase there seems 'ok' to me, suggested configs and/or default configs on those platforms may even have it enabled. For home/cpe type deployments maybe it makes more sense to have this enabled, so linksys/dlink/buffalo/etc could choose to just enable it, worst case the consumers revolt and .... it gets disabled in the next rev. -chris -------------------------------------------------------------------- IETF IPv6 working group mailing list [email protected] Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 -------------------------------------------------------------------- -------------------------------------------------------------------- IETF IPv6 working group mailing list [email protected] Administrative Requests: https://www.ietf.org/mailman/listinfo/ipv6 --------------------------------------------------------------------
