[ih] BGP vs EGP

[David Mills]

Dave,

The switch to BGP had nothing to do with manual/automatic addition of 
backbones or local nets. New backbones and local nets happened all the 
time with EGP. Nobody asked anybody, just fired up EGP and/or RIP and/or 
Hello and announced their presence. The routing protocols took care of 
spanning trees automatically as best they could. There were about 1500 
nets at the time BGP lit up. Yes, this broke the third-party rules, but 
there was no way nor no need at the time to stop newcomers. The only 
thing BGP added to the mix was a rigid administrative control and the 
resulting loop prevention.This issue was discussed in depth at the 
Technical History of  the Internet Symposium at Harvard some years ago. 
My briefing slides used at the symposium are at 
www.eecis.udel.edu/~mills/colloq.html.

Dave

Dave CROCKER wrote:

>
>
> Jack Haverty wrote:
>
>> Bob's recollections agree with mine.
>
>
>
>
> Thanks, folks.  This detail has been helpful.  Glad it's now part of 
> this list's archive.
>
> Just to summarize what I think got said:  I had understood that the 
> operational net could only support one backbone.  What I am now 
> understanding is that the issue was manual vs. automated addition of 
> backbone /routes/ and that BGP made the latter possible.
>
> d/
>