Ascend Archive
[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: (ASCEND) Ascend Max 4004 & Series56 Digital Modem Remote Gateway Probs (fwd)
Philip,
I have some experience of OSPF and EIGRP.
The advantage of OSPF are that it is an industry standard and uses link
state.
It does however require careful design as any hierachical system does
(X.400, X500, SNMP, domain names, etc). For those that do not know the
principals, it is designed in the classic upside down tree format. The
domains are aclled Areas and the root is called Area Zero. All traffic
between areas must traverse Area Zero. You could hang area 10 off Area
3 which hangs off Area Zero (for some geographic or political reason).
However all traffic between 10 and 3 would traverse 3 to zero and back.
The official way to link domains structures is to use a OSPF standards
routing protocol IS-IS. Howver in my experience BGP 4 is more common
and easier to setup.
OSPF can be frightening but if set up well is much better than any of
the older protocols and is highly interroperable. The CPU and memory
resources required mya be beyond some manufacturers current abilities -
As Some Certain Emergent Networking Devices mighth already have
discovered (IMHO) ;-/.
The only serious counter to OSPF is EIGRP. Considering Cisco is the 'de
facto' industry standard, interroperability is often not a problem.
IGRP's main weakness was it kept separate tables for each protocol and
was distance vector (hop count) orientated. EIGRP is a highly tuned
version of IGRP that shares the routing table between protocols. this
from Cisco's stand point reduces the CPU load and reduces the memory
requirements; resulting in a faster box.
In practise there is little to choose between them. Theoretically, OSPF
is better especially in large multipath WAN environments, but EIGRP is a
doddle to setup and manage.
hope this helps
ianC
Phillip Vandry wrote:
>
> > PM RIP support is stellar - it is just RIPv1. Not doing RIPv2 is not the
> > same has having problems. Livingston/Lucent deliberately skipped RIPv2 since
> > it is inferior to OSPF, and they were working on, and then fielded a solid
> > OSPF base. They had RIPv2 running, I know they guy who wrote it one
> > afternoon as a warm up, they deliberately decided not to support it because
> > it is so inferior to OSPF. And their OSPF is simple to use and reliable.
>
> I heard some valid arguments recently against OSPF. Agreed that RIPv2 is
> probably no better (or worse), though.
>
> They were pushing EIGRP instead of OSPF (they were Cisco people). Obviously
> EIGRP is not an option for anyone who hasn't got a 100% Cisco network.
>
> The complaints against OSPF is that it forces you to have a network that is
> structured with a backbone area having the required properties of the
> OSPF backbone (zero) area, that is, every other area connects to it.
>
> Because you can only summarize routes at area boundaries, you can't just
> have a single area 0 and do away with the concept of areas if you wanted
> to, for large networks.
>
> I wonder if you think these are valid concerns regarding OSPF? I have
> only limited experience with it.
>
> -Phil
> ++ Ascend Users Mailing List ++
> To unsubscribe: send unsubscribe to ascend-users-request@bungi.com
> To get FAQ'd: <http://www.nealis.net/ascend/faq>
--
+------------------------------------------------------------------+
| Ian Cowley Network Design Consultant |
| mailto:icowley@lsil.com or mailto:yelwoci@compuserve.com |
| WWW http://ourworld.compuserve.com/homepages/yelwoci |
| |
+------------------------------------------------------------------+
++ Ascend Users Mailing List ++
To unsubscribe: send unsubscribe to ascend-users-request@bungi.com
To get FAQ'd: <http://www.nealis.net/ascend/faq>
References: