For me EIGRP is not a very good idea for a couple of reasons :
* This suppose that you'll have only Cisco routers forever and make migration
quite complex if you intend a change of router vendor.
* EIGRP routing algorithm make sub second convergence implicit except in
special case if for example you have a triangle of router, the sub second
convergence work for 2 routers only due to EIGRP internal logic (successors
and feasible successors computation).
* It's generally easier to configure EIGRP, but in some cases, this protocol is
less flexible if you want to implement things like Stub and NSSA-like
You say you are looking for the "cisco" answer. In that case, I would say EIGRP. Configuring it is quite easy. If your network is simple, the main thing you have to concern yourself with is disabling auto-summarization and liberal use of passive interfaces.
From a real world perspective, an MPLS provider might be happier to exchange OSPF routes with you. In that case, the absence of mutual protocol redistribution makes OSPF a simpler choice.
My apologies... let me give you some back ground. I am looking for some CCDA material and came across a slide for recommending enterprise routing protocols. The slide is basically a comparison chart between EIGRP and OSPF of each protocols characteristics. The characteristics are:
Use of VLSM
Multiple network layer protocol support
mixed vendor devices
simple to design.
I can go through and identify which protocol has which characteristics easy enough, but tha list one is very vauge. Simple to design? My main thought is that would be EIGRP because you don't need to configured areas and have the need for planning the areas and area border routers. I did find the borderless campus Design doc that does state EIGRP being a simple protocol.
Well, the slide I am looking at says that OSPF is more simple than EIGRP and with other docs and my own experience I just have a hard time buying that. I think its an error on the slide. Just trying to validate my thinking of the "cisco way" so I can pass this thing.
Well OSPF is very simple to design and configure, as long as you're not doing any complex things. However, in a Cisco world the preferred answer would most likely be EIGRP.
I do not think Cisco would ever create a question in this way on an exam. In case of an exam they would likely include some extra information like 'the network consists solely of Cisco devices' and 'subsecond convergence in case of failure is required'. In that case EIGRP is the answer you're seeking. In questions where they're describing a hierarchical network, mergers, hetrogenous network equipment and the like OSPF would be the answer.
I'd like to agree wtih Michael about not seeing a vague question like this on the exam, but I've seen some truly obtuse and useless quesitons on design exams. You really have to take your time reading the questions, reading the answers, and finding the best fit.
With Cisco, especially with design tracks or non-"DO" exams(like specialization/partner exams):
If you have 4 possible answers, choose 1, you will have 2 answers that are flat out wrong. 2 possible correct choices, with one being a better fit, or slight more "right"(read - The Cisco Way), then the other.
The trick is figuing out which one is more right!! Typically there is a detail in the question, or a detail in the answers, that will make that difference. If you do not pay attention to said details, it can be easy to get questions wrong. This is especially true for multiple choice, multiple answer questions.
With that said, OSPF is by no means "easier" to deploy than EIGRP. If for no other reason than having to associate each network to an area, that is an extra step that EIGRP does not require. Also, multi-area OSPF automatically becomes more complex than EIGRP, but how many people that you know run a single area of OSPF across their entire network?
To be honest, I know no single area implementations, but when comparing the two in this regard you should attempt to keep things similar and that would suggest a single area OSPF implementation.
Being required to enter somewhat longer commands doesn't necessarily make it more complex does it?
However when explaining the difference between the two to a student I would also tell them that EIGRP is simpler.
Having to take areas into account does make it more complex. EIGRP requires no such consideration, either from design or configuration.
With that said, large scale EIGRP implementations can often be more complex than similar scale OSPF implementations because areas inherently scale the deployment by pruning the database depending on area configuration. EIGRP requires that to be done manually by using stub routers or summary addresses. This can break the topology because it will actually filter routes from the topology table - OSPF will always see the topology, it filters on the export to the RIB.
We could go on for a while about details and discuss semantics, but the bottom line is, Cisco thinks EIGRP is simpler to configure, regardless of single area/multi-area OSPF. Even if it were not, the convergence alone makes EIGRP a clear winner in campus designs.
EIGRP should *NEVER* be used over the WAN in my opinion. Ever.
To make it clear, I am an OSPF guy; I have used it for years and know it better so to me, I prefer to deal with OSPF, much like I prefer Cisco over other vendors, as I know the equipment. That said, I don't believe Cisco invented a proprietary protocol to do basic routing, they are the masters of networking and would have spent billions over the years building and refining IGRP and later EIGRP. EIGRP has many more features, though by default these features are not used. Also unequal cost load balancing is available, where is it is not on OSPF, so really you need to decide on a what you need it for, it’s like saying you want 20 blade servers, and 5 SANs with 7 7200 routers, to run a 5 person single office SOHO business, you need to implement what is required for the organisation. Finally, I don’t think saying “what if” you use non-cisco devices in the future is a real concern, I think I would rather implement Cisco and EIGRP, and use the EIGRP and current skill set as a business case for staying with Cisco routers, beats having to spend weeks of your own time having to learn a new vendor because the business wants to save a few thousand dollars, as well as not having to deal with multi-vendor conflicts.
To be clear I am stating there is no doubt EIGRP is the best routing protocol, you just have to decide if it is the best for YOUR situation, and business needs.
I agree 99% of the time never use EIGRP over the WAN, unless there is a design need to use it. Typically your WAN traffic is going to head out the same interface (Unless you are multihomed, etc. yes there are exceptions). But if all traffic is going to the same single destination, use static routes. That way you do not add the overhead of routing protocols on such links. Now for OSPF vs. EIGRP.
If memory serves, one reason the best practice used to be to design a multi-area OSPF network was to control the propagation of LSAs. And yes we always want to control the propagation of LSAs (Thank you stub areas....lol).
With today's routers, for your general enterprise networks, they can handle the processing of lots of LSAs. So unless your business requires multiple areas, or you want/need to practice with multi-area OSPF, a single area OSPF design should be good for a lot of networks.
I agree EIGRP is easier to configure provided there is no redistribution(Seed metric). I also agree it has faster convergence due to DUAL and Successor/Feasible Successors.
OSPF is harder to configure mostly becasue of the required wildcard mask and areas, but if I recall it is a good idea to add the wildcard mask to your EIGRP network statements anyway. If you are looking at a hierarchical design, OSPF wins since it was designed for those networks as was IS-IS. EIGRP can be designed psudo-hierarchcally via summary routes.
So it depends on what your design requirements are.
Open standard of IETF，supported by most vendors.
Cisco owned private routing protocol，not been supported by any other vendors；is not as mature as OSPF.
Most popular IGP in the world
Only a few networks designed by EIGRP，and is getting less and less popular.
SPF algorithm fast convergence, loop free.
DUAL algorithm could be in SIA status, query could spread out the whole network.
Can build a hierarchy and scaleable network.
Can not build a hierarchy network with this protocol.
Supportive of new technology
Does not support TE.
Convergence Time Fast Very Fast
I'm not sure posting a table that is clearly made by a competitor of Cisco to disparage EIGRP is helpful in a CCDA study forum. But allow me to comment:
Standard: EIGRP has been submitted to the IETF as a standard routing protocol.
Popularity: I would love to see the source of the "information" that EIGRP is used by few networks and is getting less popular. If anything, it's the complete opposite.
Algorithm: this comparison makes no sense. Both Dijkstra and DUAL provide loop-free topologies and fast convergence. The SIA mentioned is simply a symptom of a flapping link, which would have a similar impact on OSPF.
Topology: This is more of an academic argument. While EIGRP doesn't support the same topology concept as OSPF (i.e. "Areas"), you can easily provide multi-topology support via multiple EIGRP processes.
Support for new technology: OSPF-TE is a function used solely by MPLS service providers, and has been around since 1999. And in fact, you CAN use EIGRP for MPLS TE. Either way, I'd say this category was added in an attempt to make the table look bigger.
I won't re-hash what others have written, but in general, EIGRP is easier to configure and maintain for a lot of customer networks. Having said that, OSPF is also a solid choice. For the exam, you should make sure you understand the differences in how each protocol works (DV vs SPF), and also know some of the features that vary between them (like EIGRP's support for unequal cost load-balancing).