Hi Zhang,

> 1)Why we must flush the MaxSequenceNumber LSA?

Early LSP distribution schemes (ARPANET) used the lollipop seq. number
scheme in which when the seq. number reaches a maximum value it would either
wrap around. Wrapping around means that the seq. number starts at some value
(say, 0), increases to some max value (say,n), and then goes back to 0.
Given the two seq. numbers a and b, a is considered to be less than b if
|a-b| <n/2 and a<b, or |a-b|>n/2 and a>b.

A second field known as the age of the LSP is added to each LSP. It starts
at some value and is decremented by routers as it is held in memory. When an
LSP's age reaches 0, the LSP can be considered to old, and an LSP with
nonzero age is accepted as newer *regardless* of its seq. number.

This can create a problem (ref to rfc 789 for an excellent case history)
when you have LSPs from the same source S and S's LSPs have 3 different seq.
numbers: a,b, and c where a<b<c<a !

When the network gets into such a state, there is no way for it to recover.
If any router has a stored LSP from S with the seq. number b, and if it sees
an LSP from S with seq. nos c, it will overwrite the one in the memory and
make copies of the LSP for each of its neighbors. Further, a router will
flood the LSPs in precisely the order that will cause its neighbors to
accept every one( first a will be flooded, then b, then c, then a, then b,
ad infinitum!)

Since these LSPs are not getting stored in the memory long enough for the
age field to be decremented they are never aged out!

I could go on forever describing the problem and how it was ultimately
solved but this is not the issue here (I once again recommend reading rfc
789!).

To avoid all such problems a new and an improved LSP distribution scheme
came out in which the seq number is a linear space. It starts at 0, and,
when it reaches its maximum value, no other LSP from that source will be
accepted (until the LSP times out). Thus we must flush all the
MaxSequenceNumber LSAs before introducing any new LSA instances.

'Interconnections:Bridges,Routers,Switches and Internetworking Protocols' By
Radia Perlman also touches upon the above subject.

I hope i made some sense .. please contact me in case of any
queries/doubts/explanations.

Regards,
Manav Bhatia

• Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]