Re: [Lsr] Fwd: Open issues with Dynamic Flooding

[[e-mail]]

Hi Huaimo,

>    For "add temporary flooding in a rate limited manner", can you give some details about how does the rate limit manner work for fixing a FT split?


Nodes that have adjacencies that appear to be crossing the FT partition can enable temporary flooding on that circuit. This will hopefully repair the partition.  If not, the node waits awhile, and then adds another link to the FT.  Iterate until healed.


> how does each node get the rate limit?


Rate limiting causes each node to do so ’slowly’, where the exact values for the rate limiting are implementation specific.


> Will every node add temporary flooding on a given number of links independently?


No, not every node.  Only those that have an adjacency with a node that appears to not be on the FT.


> If so, there are lots of links to be added into the FT temporarily for fixing the FT split.


Yes, because this is a distributed algorithm without any centralized coordination, you can easily imagine circumstances where there are many links that could repair the partition and thus many temporary additions could be made. This is why some feel that rate limiting is wise.


> This may cause some issues.


This may be an understatement. :-)

Tony