Cisco Support Community
cancel
Showing results for 
Search instead for 
Did you mean: 
Announcements

Welcome to Cisco Support Community. We would love to have your feedback.

For an introduction to the new site, click here. If you'd prefer to explore, try our test area to get started. And see here for current known issues.

New Member

Ospf or Eigrp?

Hey all,

I am working on a project just now and I was wondering if I could get some other opinions on it please. See below.

network.JPG

My question is what routing protocol I should use? I have been told that all routers can be Cisco, so this makes EIGRP seem the most appealing to me because the ease of use and speed compared to OSPF. It says in our brief that if there is a failure a redundant link must take over in under 10 seconds so with the use of EIGRP back up routes this would be do able I would imagine. Any other opinions would be great.

Everyone's tags (3)
14 REPLIES

Ospf or Eigrp?

If you're going all Cisco, then eigrp is definitely a good way to go. The only problem that you'll run into is if you have a device, say a firewall, that's not Cisco-based. You'll need to set statics up on that device to see your internal network. On another note, ospf is a standard and supported my many vendors that are non-Cisco. I don't think you can go wrong with either, but go with eigrp if you know you'll never have non-Cisco gear.

John

HTH, John *** Please rate all useful posts ***

Ospf or Eigrp?

EIGRP looks to be a choice when all cisco. But convergence is not the only factor to consider EIGRP, as OSPF does gives you the flexibility of tuning to achieve a sub-second of convergence.

New Member

Ospf or Eigrp?

Yeh but what I meant was that because EIGRP saves the back up route it just simply switches to that, rather than have to run Dijkstra's algorithm to create a new route like OSPF does.

Ospf or Eigrp?

You are right. But what i meant to say was "OSPF even gives a sub-second convergence". So, making a choice of routing protocol may require to consider many other factors.

1) Ease of Summarization

2) Non-Hierarchical architecture

3) Open standard vs Vendor specific protocol

4) Support for unequal cost load balancing - Only EIGRP supports them.

So, above are some things i wanted to let you know. You might have already considered them, but still wanted to highlight.

Thanks

Vivek

New Member

Ospf or Eigrp?

Thanks Vivek,

Yeh I already considered them. All the links are 2mbps so it wouldnt benefit from unequal load balancing. I haven't got any experience in using these in a real world scenario other than labs so I was just looking for information in regards to problems encountered with the protocols.

Thanks

Cisco Employee

Ospf or Eigrp?

Dear friends,

Allow me to join.

Yeh but what I meant was that because EIGRP saves the back up route it  just simply switches to that, rather than have to run Dijkstra's  algorithm to create a new route like OSPF does.

It is not at all clear that EIGRP "simply switches to a backup route". The backup route would need to go through a feasible successor and it would need to be the next least-cost path immediately after the current successor route. In all other cases, EIGRP needs to go active and perform the usual diffusing computation trick. It is quite difficult to meet the above two criteria for immediate successor changeover. In many networks, feasible successors do not exist (as the neighbors do not pass the feasible condition check), and it is also difficult making sure that the feasible successor provides the second least-cost path to a destination. In fact, the entire concept of feasible successors in EIGRP is not a primary design feature but rather a byproduct of how the feasibility condition works. It is a nice add-on but don't rely on it.

Another thing to consider with EIGRP convergence is that it may easily take longer than in OSPF because in OSPF, you can run the SPF immediately without waiting on your neighbors. EIGRP is dependent on receiving all replies from all queried neighbors before making the next step, and if this process is lagging, the convergence in EIGRP slows down - not because of the complexity of the algorithm but because of delays in getting all necessary replies back.

In a real network, you woud probably perform a series of tests with different settings to see how would the individual routing protocol perform. There is no clear answer as to which one is better, and none of them would be better on all accounts.

My two cents...

Best regards,

Peter

New Member

Ospf or Eigrp?

Thanks for that informaiton Peter.

So do you think rather than rely on a back up route, I would maybe create a floating static route?

Thanks

Cisco Employee

Ospf or Eigrp?

Tony,

I do not think that would speed up things. In order for a floating static route to kick in, first the primary route would need to be withdrawn from the routing table which means that the EIGRP or OSPF algorithm have detected the change and have completed their recomputation of the new set of shortest paths. You see, in this case, the floating static route would not be faster than the routing protocol itself.

My remark was related to comparing the speed of EIGRP and OSPF, and was intended to point out that the issue of fast convergence is more complex than just making quick judgements based on their basic properties.

Is this a real network project or more of a lab/seminal assignment?

Best regards,

Peter

New Member

Ospf or Eigrp?

Ok thanks. It is a assignment for Uni.

Ospf or Eigrp?

Hi,

IMHO,  in a small network like this the Dijkstra's  algorithm itself would run so fast it has no practical sense to consider any covergence delay caused by that.

BR,

Milan

New Member

Ospf or Eigrp?

Good point. So which one would be quickest do you think?

Ospf or Eigrp?

Thanks Peter for entering into the discussion.

Tony,

Just to add to peter's explanation. As you said "EIGRP just injects the backup route into the routing table instantly. It's true to some point. If you have a backup route in topology table which you see it to be feasible successor, then yes EIGRP injects it. But there are some situations where the backup route isn't a feasible successor. Remember ? ADFD route (no choice).

I tried explaining this in my way

Cisco Employee

Ospf or Eigrp?

Hello Vivek,

Thank you!

One remark: you are using the term "FD of succesor". There is no such thing. The Feasible Distance, or FD, is a parameter of the route itself, and is not related to any particular neighbor or successor. It is a common mistake in textbooks which talk about "FD through a particular neighbor" or "FD through the current successor".

The FD is defined as the smallest distance to the destination since the last time the route entered the Passive state. In other words, it is a record of the historical minimum of the distance towards a particular destination. FD is not the current best distance, and it is not related to any particular neighbor. It is simply a local variable holding the minimum value of the distance towards a particular destination - and moreover, it is never advertised. It is entirely a local parameter.

So the feasibility condition should better be described as "AD of the neighbor < FD". Personally, I even replace the "AD" with "RD" (reported distance), because the "AD" as advertised distance tends to be confused with "administrative distance".

Best regards,

Peter

Ospf or Eigrp?

Hi,

I personally prefer OSPF.

IMHO, there are some vulnerabilities in EIGRP (see

http://www.cisco.com/en/US/tech/tk365/technologies_tech_note09186a008010f016.shtml

http://www.cisco.com/en/US/technologies/tk648/tk365/technologies_white_paper0900aecd8023df6f.pdf  )

which could bring you in trouble in some cases.

But that's a question of personal preferences,  I guess.

BR,

Milan

2702
Views
19
Helpful
14
Replies