Messages in this thread | | | From | "Fred Reimer" <> | Subject | RE: howto disable auto route setup? (Delete linux-kernel if followup) | Date | Wed, 3 Feb 1999 15:03:53 -0500 |
| |
> -----Original Message----- > From: owner-linux-kernel@vger.rutgers.edu > [mailto:owner-linux-kernel@vger.rutgers.edu]On Behalf Of Vijay G > Bharadwaj > Sent: Wednesday, February 03, 1999 11:52 AM > To: Fred Reimer > Cc: linux-kernel@vger.rutgers.edu > Subject: RE: howto disable auto route setup? > > On Tue, 2 Feb 1999, Fred Reimer wrote: > > Show me the RFC where it says this is legal all of a sudden.
> RFC 1518 and 1519. Draft standards. Dated September 1993, not all that > recent either. 1519 even has an example with a network mask of > 255.255.255.240. > > A/B/C addressing is dead, and has been for a long time. Support for it > only remains to support legacy configurations. > > Please let's kill this thread.
O.K., this thread, as far as CIDR, et. al. is dead from linux-kernel. If you want to reply to one of my emails, make it private. I'd be glad to summarize to the list if anyone is interested after it gets hashed out, but this does not belong on linux-kernel.
Vijay,
RFC1518 specifically states that it does not address and makes no recommendations on the assignment of IP addresses/masks to hosts. CIDR is a ROUTING concept. It was never meant to be used for host IP address/mask assignments. It was meant to lighten or prevent increasing the routing table load on Internet backbone routers. I agree with it. It's great! It's old. Been there, done that. Besides, 255.255.255.240 is a "legal" subnet on any (A/B/C) network :-) You have always been able to subnet "more" than the network class. My argument is that it's not documented anywhere that you can subnet "less" than the network class for >an interface<. You have long been able to do this for >route< aggregation, as so many people have pointed out.
This is the first time in over 7 years of my professional network consulting career that I have heard of anyone using it on an >interface<. It IS a Good Idea(tm). I wish it was available when I was setting up networks years ago. Sadly, it is NOT "available" to me because all of the clients(hosts) that I have to deal with do not support it. Why not? Because no one wrote up an RFC and sumbitted it (not even Informational) that says hosts should support it. RFC 1618 does indicate it would be nice, but this is an Informational RFC and AFAIK was never followed up with an Experimental one: "Merging Subnets
There has long been some need to merge subnets. Sometimes this is due to organizational changes; other times, people have installed bridges when routers would have been a more appropriate choice. Some
Bellovin [Page 4]
RFC 1681 On Many Addresses per Host August 1994
hosts need to live on both logical networks at once, to avoid an extra hop through a router. It would be useful to be able to assign them such addresses. "
RFC2101, which I just re-read and is dated just two years ago at February 1997, does not mention host interfaces using CIDR like concepts. It's all routing issues and aggregrating routes for Internet routers. AFAIK, there has been no update. If anyone can show me otherwise, feel free to reply. I have searched the RFC's, even reading every title after 2101, and could not find anything. I'll humbly appologize for the waste of bandwidth and promise to read every RFC from 1 to 2509 to make sure I'm up on everything if I'm wrong.
Note to everyone: although it may seem like it, I am not arguing for the sake of arguing. It's my job to determine IP addressing schemes as part of installing networks. If I have been missing out on something that has been available for years, I'll be mighty pissed (at myself). However, if it's not a documented standard, and if any of Microsoft's OSs (3.11, 95, 98, NT3.51, NT4.0) do no support it then it's not an option for me "professionally" to use it. You can use it in your own network that you administer, but networks I install are there years after I leave the site and who knows who will be administering them in the future? Plus, they all use M$ products, which I can not force them to change (as much as I'd like to). I do support the notion, and think it's a good idea though. I would support making it a standard (or Standard if you want to nit, but it should at least be on-track before "I" start using it professionally).
I'd be happy to work with anyone to write up an RFC and submit it for consideration if you are interested. I, frankly, don't feel comfortable doing it myself because I have never seen and experienced a site myself that does use it.
Again, DELETE linux-kernel from the header, please.
Fred
- To unsubscribe from this list: send the line "unsubscribe linux-kernel" in the body of a message to majordomo@vger.rutgers.edu Please read the FAQ at http://www.tux.org/lkml/
| |