At 01:56 PM 2/15/02, Steven A. Ridder wrote: >Are you saying that the fact the router's routing process that sent the >packet to be forwarded considers the packet "gone" even though it's still in >output queue's buffer being transmitted?
Yes. That would be good router architecture. A paper I found on CEF has some nice drawings of router "modules." See that the IP Input Process that does the forwarding is separate from the Output Interface Processor in the drawing here: http://www.cisco.com/warp/public/105/20.html This discussion does point out an interesting thing about TTL. It never really was a time to live on the internetwork, because it didn't take either serialization or propagation delay into account. It just took router processing delay into account. I think that's because the router processing time was an issue back in the 70s and 80s. Serialization delay also contributed to overall delay, of course, but wasn't as big an issue as processing delay. (I think the original ARPANET was built on 9600 bps lines.) Propagation delay hasn't changed much over the years and isn't a big contributor compared to the others, (depending on distance). Bits still traveled at approximately 2/3rds the speed of light back then just like they do now. But the Honeywell Interface Message Processors (IMPs) that acted as routers were the size of a refrigerator, buggy, sluggish, and hard to program. I don't know MIPs or MHz, but we're talking SLOW. ;-) Priscilla >-- >RFC 1149 Compliant. > > >""Priscilla Oppenheimer"" wrote in message >[EMAIL PROTECTED]">news:[EMAIL PROTECTED]... > > At 01:25 PM 2/15/02, Hire, Ejay wrote: > > >I lab-ed this, and did not observe the TTL incrementing even when the >delay > > >was over 8,000 ms. (It's not how fast you send the packets, but how slow > > >you make the link!) > > > > Hmm, that's an interesting approach, but I'm not sure it's a valid test. > > Think about the layering and modularization of protocols and router tasks. > > IP forwarding doesn't know how long it takes to output bits. It couldn't > > decrement the TTL based on the delay in sending, even if the TTL really >did > > still have a time-based meaning rather than a hop count meaning. > > > > Priscilla > > > > > > > > >-----Original Message----- > > >From: Michael Williams [mailto:[EMAIL PROTECTED]] > > >Sent: Friday, February 15, 2002 11:54 AM > > >To: [EMAIL PROTECTED] > > >Subject: RE: TTL and modern (fast) routers [7:35507] > > > > > > > > >AFAIK, the TTL gets decremented by one by a router as it passes it on (if > > >it's held under one second), or by the number of seconds it was held if >it > > >is held over one second. I agree that anything more than 1000ms of delay > > >seems outrageous for a single hop these days, but I don't know of >anything > > >that has changed that "rule" that both you and I describe. > > > > > >Mike W. > > ________________________ > > > > Priscilla Oppenheimer > > http://www.priscilla.com ________________________ Priscilla Oppenheimer http://www.priscilla.com Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=35536&t=35507 -------------------------------------------------- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
