First, from the certification perspective, you don't really need to
know the details of the link state route computation algorithm.
Second, Jeff Doyle and Radia Perlman wrote fairly generic
descriptions of what is happening, but not at a detail of precision
that is needed to implement the code. RFC2328, although no paragon
of clear writing, is at that level. John Moy's _OSPF: Anatomy of an
IP Routing Protocol_ doesn't go into the computational details,
although his new one, _OSPF Complete Implementation_, appears to do
so (I just got it).
Third, there are differences in the conceptual algorithm and the way
it is implemented. See http://www.nanog.org/mtg-0010/igp.html for a
discussion of concerns before we can reach millisecond convergence in
interior routing protocols. Briefly, Dijkstra's algorithm is 40
years old, and there are alternatives that produce the same output
but much faster.
Fourth, route computation algorithms, in general, really are complex.
You won't understand them thoroughly without a good knowledge of data
structures and search algorithms, typically an upper undergraduate or
graduate course in computer science. Thorough understanding of
finite state machines is essential. You will also need to track the
IETF working groups to understand implementation issues.
Believe me, you don't necessarily need to get into all the details.
I'm the routing software architect in my product development group,
and on the one hand need to restrain my hardware colleagues into
getting too deep into algorithms that will be in software, and on the
other hand remember that I may not be the person to write the fastest
code (any more...sigh).
>"Katson PN Yeung" <[EMAIL PROTECTED]> said,
>
>For me, I find Radia Pearlman one is easy to follow....
>
>
>"Jaeheon Yoo" <[EMAIL PROTECTED]> wrote in message
>[EMAIL PROTECTED]">news:[EMAIL PROTECTED]...
>> Hi,
>> Merry Christmas and Happy New Year to All!
>>
>> I'm reading RFC 2328, fortunately most part of it is understandable
>> for me. But "16. Calculation of the routing table" part is extremely
>> hard to follow. Although Jeff Doyle's explanation of Dijkstra's
>> algorithm sounds easy and interesting, why is it so tough in original
>> document? Is this essential or required for me to thoroughly
>> understand it in its original terms? In that case, are there any plain
>> language version of it out there?
>>
> > Thanks in advance
_________________________________
FAQ, list archives, and subscription info: http://www.groupstudy.com/list/cisco.html
Report misconduct and Nondisclosure violations to [EMAIL PROTECTED]
