[c-nsp] TCAM refresher

[Tony Li]

On Mar 17, 2007, at 5:45 PM, Chris Woodfield wrote:

> Tony, which type of aggregation did you benchmark - adjacent prefix
> aggregation (x.0/30 + x.4/30 = x.0/29), or overlapping prefix removal
> (x.0/24 + x.128/25 = x.0/24)? If both, which one seemed "easier" to
> do? Do you recall what sort of FIB size savings were realized?


Chris,

We did both of those, plus several other variants.  None of them is  
particularly hard by itself, but being sure to track changes if one  
of the involved prefixes was definitely painful.  Note that I really  
can't take credit for this as other good folks did the  
implementation.  I'm not sure that they're looking for fame, glory  
and the concomitant spam, so I'll avoid mentioning them by name.   
I'll just say they took a nasty problem and did a great job.

For realistic situations, we would see a typical case improvement of  
about 30% of the size of the table.  Worst case was about 10%, best  
case was about 45%.  The dominant factor in the variance was the  
diversity of next hops that a particular box used.

Tony