>One thing to remember is that OSPF costs are calculated >unidirectionally. For example, A's cost to C could be very different from >C's cost to A. In general, IP traffic has to be engineered in both >directions and it for some networks asymmetry in flow might make sense. I >can't think of a reason off hand while watching a hockey game, but >experience has taught me that many odd looking designs are rooted in >rational, informed theory. > >Pete
There's a whole class of applications for this sort of thing, and indeed there is a Unidirectional Link Routing group in the IETF for dealing with them. Much of the leading research work is in France, where practical applications deal with using high-bandwidth satellite links in one direction to deliver educational content to Africa, and a low-speed terrestrial return link for acknowledgements. ----- Following up on this general issue of asymmetrical satellite+terrestrial communications, you might find this interesting: At 8:42 AM +0200 3/5/02, Hank Nussbacher wrote: > >New 12.2(8)T feature in Cisco IOS called TCP Windows Scaling: >http://www.cisco.com/univercd/cc/td/doc/product/software/ios122/122newft/122t/122t8/tcpwslfn.htm > >Specifically made for satellite networks: >ip tcp window-size 750000 > >-Hank In general, TCP needs help to deal with long-delay or very-high-speed media. There is an assortment of RFCs on this, but this is more a matter of host than router concern. As a rule of thumb, you start getting in trouble with standard TCP at about 50 Mbps, and run into theoretical limits at about 600 Mbps. There are workarounds such as extended window and sequence fields, plus selective acknowledgement. Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=37265&t=37265 -------------------------------------------------- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
