EIGRP neighbors

Unanswered Question
Mar 16th, 2004
User Badges:
  • Cisco Employee,

I am looking for documentation on EIGRP scalability. I need to know what the recommended number of EIGRP neighbors would be for a 6500 or other platforms (a matrix would be nice). I am wondering when I should use passive interfaces to reduce the size of the neighbor table and is too much redundancy in EIGRP a bad thing? thanks for the help.

  • 1
  • 2
  • 3
  • 4
  • 5
Overall Rating: 0 (0 ratings)
ruwhite Tue, 03/16/2004 - 13:58
User Badges:
  • Gold, 750 points or more

It depends. :-)

I say that tongue in cheek, but it really is true, it really depends on a lot of factors.

-- How many routes are you sending to each neighbor?

-- Are they stubs, or normal remotes?

-- Are they dual homed, or single homed?

A matrix would be nice, but it doesn't exist, and probably won't ever, because people will tend to take it as "gospel," and build their networks around it, which can result in blown up networks and people pointing fingers, which is never any fun. :-( If we list optimum condition neighbor counts, those without optimum conditions will complain that they can't reach those numbers, and if we list worst case numbers, those with optimum conditions would complain that we're underutilizing the hardware. It's too nuanced in the real world....

I have seen networks as high as 600 neighbors, ideal conditions, working well, and networks as low as 50 or 60 neighbors fail miserably. As an internal person, you can email me, and I can point you to the studies on scaling we have available, and work with you on specific network design issues. :-)

Too much redundancy in EIGRP is a bad thing, in general. I wouldn't recommend more than the number of links set in max paths, since that's where we tend to start hitting problems.




This Discussion