I have one more comment about reliability and load. As I stated before, they are dynamically calculated. It made some sense to take them into account with IGRP because of the regular updates every 90 seconds. A recipient router could maintain a somewhat real-time view of the reliability and load for a path and theoretically adjust to changes (select a path with better reliability and load). EIGRP only sends updates when there's a change, however. By then it's usually too late to be telling the other side about changes in reliability and load. The result could be pretty chaotic.
Howard has discussed modern routing protocol principles in the past, so the archives probably have some good stuff on this. Regarding MTU, I've never gotten a straight answer on what the MTU is really used for in IGRP and EIGRP. We know it's not technically part of the calculation of the composite metric because of what we've learned about the K equation. But we do know that routers spread the info about MTU and you can see that they save info about it when you do a show ip eigrp top command. Is it just FYI?? I've read tons of material on EIGRP and never found an answer to this question that pops up every so often. Maybe you could do some lab testing? :-} I'm buried in other types of problems just now and can't do it. But what happens if you have two routes to a network that are both equal except for the MTU? Is the one with the higher MTU chosen?? Priscilla At 12:09 AM 4/21/02, Michael L. Williams wrote: >coming to check Groupstudy, I was going though a CCIE practice exam, and it >asks "What's used by EIGRP to determine the best path?" and for the answer >it lists all five items, not just the four (that we all agree are used in >metric computation). Now, given that we know the formula for the EIGRP >metric (that doesn't include MTU), and given the recent thread about which >routes are used in the routing table (i.e. routes with the longest mask, >lowest AD, etc), would you agree that the above sample question is >incorrect, or does EIGRP actually use MTU to aid in determining the best >path? If so, how/where? I'd like to know so I can get on to other issues >like MPLS, etc before I take this beta =) > >Regarding my comment about the recalculations related to K1-K5, I got it >now...... At first I was like "Why would there by any more recalcs than >normal", but as Kevin pointed out, load, etc can vary over time, whereas >bandwidth and delay are constant. Second, I was thinking "Okay, then I can >see that causing recalcs that affect the routing table, but the topology >table?", so I hopped into an EIGRP router and with "show ip eigrp top" saw >where I wasn't thinking. The topology table not only keeps the metric for a >given route and it's feasible successor, etc so yeah... now I get it =) > >I'm slow sometimes, but I usually come around =) > >Mike W. > >"Priscilla Oppenheimer" wrote in message >[EMAIL PROTECTED]">news:[EMAIL PROTECTED]... > > Reliability and load are dynamically calculated. They change frequently. > > Their changes could cause the routing table to change too, maybe more than > > you would want. Delay and bandwidth are not dynamically calculated. > > >"Kevin Cullimore" wrote in message >[EMAIL PROTECTED]">news:[EMAIL PROTECTED]... > > I'd expect that the values they are multiplied by would change if they are > > truly based upon data derived from the interface counters, which are > > certainly subject to potentially vast fluctuations in absence of topology > > changes ________________________ Priscilla Oppenheimer http://www.priscilla.com Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=42182&t=42103 -------------------------------------------------- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
