Just for clarity, this really isn't a split horizon implementation. Specifically, split horizon deals with prefix advertisements directed out of interfaces from which they were originally learned which is not what IBGP does. IBGP simply does not re advertise prefixes learned from IBGP neighbors, irrespective of interface.
Although I agree with your math on full mesh peering, I'm not clear on the your presented negative opinion wrt to route reflection? Peering requirements are dramatically reduced due to route reflector clustering and route reflection is almost universally utilized in BGP networks. Pete At 01:06 PM 1/6/2002 -0500, Mike McCline wrote: >Hi Vincent >The Split Horizon concept as it applies to distance verctor protocols is >used with BGP. Routes learned via IBGP are never propagated to other IBGP >peers. To alleviate this issue you have two implementation choices, use a >full-mesh IBGP setup or route-reflectors. For full mesh the number of >sessions can be calculated using x=n(n-1)/2 so for a 1000 routers you would >nearly half a million sessions. On the other hand if you use route >reflectors to solve the issue of split-horizon the number of sessions would >be in the thousands due to clustering. Check CCO for a more detailed >explanation. > > > >--------------------------------- >Do You Yahoo!? >Send FREE video emails in Yahoo! Mail. Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=31182&t=31063 -------------------------------------------------- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]