Eigrp unequal cost load balancing

[haris puthiyottil]

Hello,

I am planning to do unequal cost load balancing using eigrp . I have below setup.

Core(6500)-------(router1)---4MB---(WAN)       

      \

        \--------------(router2)-----6MB--(WAN)   

1)Here I am planning to configure variance only to the nework connected to router 1 and router 2. If I configure variance 2 under eigrp in Core switch that will affect all the network under the same eigrp AS.Is there any way to apply variance only on the network which is connected to router 1 and 2

2)Eigrp will consume 50% of bandwidth by default. Is it only eigrp control traffic.

[JohnTylerPearce]

As far as I know, the 'variance x' command, is a global command, that can't use an ACL or anything like that.

If you want to use both router on the Core(6500) which are going out R1 and R2, you could use an 'offset-list', and make sure the routes going to R1 and R2 are equal. Just a thought.

[haris puthiyottil]

Changing metric can be acheived by making changes in delay. But I am not planning to do equal load balancing since I have 10Mbps and 6 Mpbps link and doing so make 6Mpbps link over utilized. Hence I am thinkg to go for unequal load balancing.

second question what all trafiic comes under default 50% eigrp traffic

Appreciate your response

[JohnTylerPearce]

Well, EIGRP is a pretty quite protocol, it's not like OSPF. Once a neighborship is established, it will push out its full route table, and then only updates after that. Of course, you can have a active/passive situation as well. You can also change the percentage that EIGRP uses for traffic, you can do this by using the 'ip bandwidth-percent' command.

Now what you could do is.....

Create a different EIGRP AS on the Core(6500) and then you could enter the variance command, which would only affect the routes going to those two routers, but you could have to change the AS on R1 and R2, and then think about redistribution, since I"m sure you want full network connectivity.

[Peter Paluch]

John,

Just a small nitpick:

Well, EIGRP is a pretty quite protocol, it's not like OSPF

OSPF is quiet, too. After the LSDB are synchronized, routers reoriginate their own LSAs roughly each 30-50 minutes. While OSPF is more chatty in this respect, the consumption of bandwidth by OSPF is negligible, and I would not put that in contrast to EIGRP.

Best regards,

Peter

[JohnTylerPearce]

Peter, nice to hear from you! I always like it when you nitpick peter, id rather not say something stupid if I can avoid it, so please feel free at any point.

While I agree that once OSPF is configured, and LSDB are exchanged, they will reoriginate their own LSAs, like you said, every 30 to 50 minutes, since you have the 1 hour cut off time, this still happens every hour. Also, if you have any sort of link change in OSPF, it would flood this change, throughout the entire AREA. As well as every 10 minutes, performed a periodic check-summing and aging procedure and flush any non self-originated aged and or corrupt LSAs.

With EIGRP, it's basically configured, setup, and you don't have to worry about periodic flooding etc. The only think you will really have to worry about, is if a route goes ACTIVE and end's up getting SIA.

Now this also depends on how much routes you have, etc etc.

[Peter Paluch]

Hello John,

Oh, it is always nice to talk to you, my friend.

As to the chattiness of OSPF versus EIGRP... EIGRP is a quieter protocol for most operations, that's for sure. But I felt that OSPF was put into a disfavorable position while in fact, despite it is more chatty, the amount of information flooded by OSPF is - in my opinion - hardly a reason for concern. Yes, it does send information infrequently even if the topology is stable. The query/reply process in EIGRP, on the other hand, can be more intensive with regard to messaging that a simple updated LSA flooding.

As well as every 10 minutes, performed a periodic check-summing and  aging procedure and flush any non self-originated aged and or corrupt  LSAs.

I am sorry - what did you mean by those 10 minutes? I am not aware of that interval in OSPF.

It is true that one of the most serious flaws to OSPF's scalability is the standalone nature of individual LSAs. Each of them has its own ID, its own lifetime, can be individually requested, updated, transmitted, has to be acknowledged, refreshed or expired, and there can easily be thousands of them in a single area.

With EIGRP, it's basically configured, setup, and you don't have to  worry about periodic flooding etc. The only think you will really have  to worry about, is if a route goes ACTIVE and end's up getting SIA.

I would be very careful with this complacency towards EIGRP It has its own quirks, and one of the most dangerous things about it is the perceived easiness of deploying it without a true need to know or understand how it works inside. We have a saying in Slovakia that could be roughly translated as "Quiet waters erode the banks."

Best regards,

Peter

[JohnTylerPearce]

The way Cisco routers originally did that was by running a refresh procedure every 30 minutes and refreshing every self-originated LSA in the database, no matter how old it was. This would result in sudden CPU spikes every 30 minutes in case of large databases in addition to bursty LSA flooding. Every router in the routing domain in turn would have to receive and process a large amount of LSA information. This is a good example of a “synchronization” problem. In addition to this refreshing, every 10 minutes a router would run periodic check-summing and an aging procedure, and flush any aged non-self-originated or corrupted LSAs.

The above is from INE.