From: Howard C. Berkowitz (hcb@gettcomm.com)
Date: Sat Oct 12 2002 - 12:29:45 GMT-3
Peter van Oene commented,
>At 08:40 PM 10/12/2002 +1000, Nick Shah wrote:
>>Brad
>>
>>No one protocol is better than the other. Both have their Pros, and since
>>they are highly mature protocols, they really dont have that many cons. A
>>few facts related to both of them, from an operational perspective ...
>>
>>- OSPF is more CPU intensive than EIGRP
>
>Modern routers have little trouble with either
>
>>- EIGRP is a Cisco proprietory protocol, so if you are using non-cisco
>>routers/devices in your network OSPF is your only choice.
>
>I would suggest that this is a big deal.
>
>>- MPLS capability, if desired, leaves you only OSPF as the choice (and ISIS,
>>of course). I had heard that there was some work happening with making EIGRP
>>mpls capable (the TLV functionality is currently absent).
>
>Of course this only relates to MPLS-TE, not MPLS in general.
>
>>- OSPF requires more management (even for medium sized networks), EIGRP is
>>straightforward for medium sized
>>networks (less mgmt. intensive)
>
>I might agree that it is easier to setup a medium sized network with
>less clue, but clue is a good thing for decent sized networks and
>clueful EIGRP and OSPF networks likely require similar amounts of
>effort.
>
>>- OSPF has no scalability limits, with EIGRP you really have to do proper
>>summarization & stub routing etc. to achieve the scalability of OSPF (so as
>>network size grows, you will find that the complexity working with EIGRP
>>increases)
>
>Neither protocol has limitless scale, however, OSPF has more
>inherent scaling tools. One can achieve similar ends with EIGRP,
>however eventually one tends to run into SIA issues as you decribe
>below.
>
>>- EIGRP has faster convergence than OSPF (very slightly, but yet it beats
>>OSPF at speed)
>
>This is a very contestable item. Various scenarios can be designed
>with both protocols to highlight strengths and weaknesses.
Both are fast. EIGRP will be faster if the alternate route is
present in a neighbor. OSPF will be faster if the alternate route
needs information from a router more than one hop away.
This assumes, of course, that the hello timer is set identically.
Its default value in EIGRP is 5 seconds but 10 seconds in OSPF, so
unmodified EIGRP certainly will detect failures faster.
>
>>Now a couple of cons..
>>
>>- OSPF's LSA flooding can sometimes be too much for a lesser powered router
>>and low bandwidth links, well so is EIGRP SIA (stuck in active). However in
>>a well laid out network (use stub areas for OSPF , summarization boundaries&
>>stub routing in EIGRP) thats not too much of a problem.
>>- Area mgmt. with OSPF can be quite cumbersome, however with a proper
>>design, the complexity is paid off with increased stability.
>
>I'm not sure I'd agree that area management is cumbersome.
As Peter alluded, at some point, you have to have significant clue to
run a good-sized network. I've done networks with 2000-3000 routers,
although most were edge routers statically routed to OSPF
distribution routers. Putting OSPF only on the routers that have
enough alternate connectivity to make use of it is a BIG saving.
>
>>So if your or customers network is still in design stage, you can weigh the
>>pros & cons and decide for urself.
>>
>>Thanks
>>Nick
>>
>>
>>----- Original Message -----
>>From: Hung, Sing-Yu <Sing-Yu.Hung@pccw.com>
>>To: <ccielab@groupstudy.com>
>>Sent: Saturday, October 12, 2002 4:48 PM
>>Subject: OSPF Or EIGRP
>>
>>
>>> Dear,
>>>
>>> I just want to know which protocol is better? OSPF or EIGRP. Because
>>> my client want to change from OSPF to EIGRP without any reason. And May I
>>> have your suggestion?
>>>
>>> Bradford Hung
>>>
>>> Pacific Century CyberWorks
>>> Tel: 288 33125
This archive was generated by hypermail 2.1.4 : Tue Nov 05 2002 - 08:35:45 GMT-3