At 7:06 PM +0000 7/21/03, alaerte Vidali wrote: >Thanks, > >The challenge I am facing is to improve the OSPF design of a network that is >in production, without changing the existing WAN links. The area 5 is bigger >than I think it would be ideal (there are 56 routers) and there is no >interesting traffic between all the spoke routers in the area, as Howard >mentioned.
Let's come back to basics. Are there actual problems with the current design, or are you simply concerned about an arbitrary number of routers? One of the important things is to look at the processor utilization of the hub router. If the five-minute average isn't much beyond 50-60%, you probably aren't having any real problems. Admittedly, if the whole area goes down and then up at once, you may have some peak load issues. But in networks of reasonable stability, the RISC processors let you have more routers than the old 50 or so limit. These days, I'm usually more interested in avoiding a single point of failure than I am squeezing out the last bit of performance. > >In the beginning I were wondering about transforming R1 in an ABR; but I >gave up because the stability of the backbone (R1 is not so powerful as the >others, R2 and R3). Also, today there are two ABRs in area 5, and it would >have just 1 ABR with this change. >Now I am considering to change the spoke routers connected to R3 to a new >area. Today there is only one exit point for the spoke routers connected to >R3, and it would not be changed. > >R3 would be connected to R2 (through area 0 and area 5) and to R1 (through >area 5), as it is today. >R2 and R3 would be the ABR for area 5; no changes here. >R1 has an ATM link to its two ABRs, R2 and R3. The main link is to R2; the >link to R3 is a backup. >R3 would participate in area 5 just because the existing WAN link to R1. >R3 would be the ABR for the new area, probably area 6. >The benefit is a small database for area 5 and area 6, without flooding >information where it is not useful. > > >Is there a document explaining the rules of summarization when there are two >ABRs on a area? I discuss the problem in several of my books, but it's more an issue of fundamental address design than it is summarization per se. The major issue is the tradeoff between stability and reachability, which calls for different levels of summary assuming a systematic address assignment plan. Message Posted at: http://www.groupstudy.com/form/read.php?f=7&i=72747&t=72587 -------------------------------------------------- FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
