[Cu-wireless] Re: Are we making things too hard on ourselves?

[David Young]

Ben,

There are several protocols out there for broadcasting/computing
network routes. Examples are RIP, BGP, and OSPF.

I do not understand too well what these protocols do, but here I take
a stab at explaining some of them.

Ordinarily, BGP (Border Gateway Protocol) is used to compute routes
between entities called "autonomous systems" (AS's). An autonomous
system is ordinarily some large network that it is not necessary or
desirable to conceive of as something for you to route your packets
through. The University, with all its departmental networks and its
complicated network infrastructure, is a good example of AS. So, too,
is any university, or a major corporation and/or its offices.

OSPF is run inside an autonomous system to compute routes from department
to department.

At any autonomous system, no more than a few routers will usually run
BGP. These are the routers that serve as gateways to the rest of the
Internet. As far as the network topology is concerned, they are located
on the "borders" of the autonomous system. Hence Border Gateway Protocol.
Usually these routers will run OSPF, too, which keeps routes to networks
inside the AS and also to border gateways.

Most the other routers in an AS will run OSPF.

RIP is a really ancient protocol. It is still useful, though.

I don't think any router usually chooses protocols dynamically. Network
conditions probably do not change so drastically that where BGP was
suitable, OSPF will be.

A few more words about these protocols.  OSPF is called a link-state
protocol. It forms a (weighted) connectivity graph and computes a
spanning tree.  The tree basically dictates what routes are chosen.

BGP is called a distance vectors protocol.

Dave

On Fri, Jan 18, 2002 at 04:26:51PM -0600, ben emerick wrote:
> hi paul, 
> your explanation was helpful.
> i'm new to the wireless scene and old computers.  could you or anyone else 
> go into more detail about different protocols which a router might use to 
> decide which of several possible paths to use for a particular packet?  are 
> there already available implementations which can change protocol, based on 
> network monitoring, without user intervention?
> if this sounds too long for email, maybe someone would be willing to talk for 
> a few minutes at the meeting this weekend.  
> 
> ben
> 
> * Paul Kennedy <pkennedy at nplldt.npl.uiuc.edu> [18.01.02 14:24] stated:
> > 
> > [
> >     I get the digest of this list, so I don't know whether anyone
> >     has talked about this before.  Apologies if this has already
> >     been covered.
> > ]
> > 
> > niteshad at whopper.de wrote:
> > > I was just thinking about our conversation at the last
> > > meeting regarding ways to get cheap wireless routers,
> > > which ultimately terminated in the decision to look into
> > > the University Auction in Springfield.  We decided that we
> > > needed something with an 802.11b card, an RJ-45 Ethernet port
> > > and the ability to do routing, well, isn't that exactly the
> > > definition of a Wireless Access Point?
> > 
> > Unfortunately, no.  Here's part of the difference.  In most
> > "apartment network" setups, the only thing that needs to be
> > done is "routing" like this:
> > 
> >        [Node A]-| ....  
> >                        .
> >        [Node B]-|  .... . |-[ HOST Z ]---->[modem]----[Internet at large]
> >                        .
> >        [Node C]-| ....  
> > 
> > 
> > Host Z acts as a fairly rudimentary router (and possibly Network
> > Address Translation, or NAT box).  In this simple network, the
> > decisions that host Z has to make are fairly simple: did the
> > transmission come in on the wireless port side?  does this next
> > packet need to go out on the wireless side or the wired side?
> > Host Z is really acting as a gateway.
> > 
> > Consider the mishmash below (dashed lines represent links,
> > wired or wireless, it doesn't really matter).
> > 
> >                        B-----C-----D---[Internet]
> >                       /      |      \
> >                      /       |       \ 
> >     [Internet]---- A         E--------F
> >                      \       |       / 
> >                       \      |      /
> >                        G-----H-----J---[Internet]
> > 
> > Nodes A, D and J have links to the Internet.  Where do packets
> > coming from E go?  Do they go E-C-D or E-F-J or E-H-J or E-F-D?
> > With real routers, we could pick amongst various routing
> > protocols to find what works best for the collection of nodes
> > that we'll end up having (as well as being resilient in the
> > face of node/link failure).
> > 
> > As a side note to the network junkies, my guess is that we'll
> > probably end up running something related to OSPF since it is a
> > link state protocol (instead of a distance vector protocol with
> > their associated problems).  I have no idea of the consensus
> > of the group.
> > 
> > So the real problem with trying to use just a wireless access
> > point is that you are most likely _not_ getting a true router.
> > A true router should be able to deal with network loops (as
> > in the case above) as well as deal with things like nodes or
> > links going down (as might happen with a community network).
> > 
> > I hope that I've not caused any confusion and hopefully even
> > helped :)
> > 
> > Paul
> > 
> > 
> > -- 
> > Paul A. Kennedy    "I don't use VI unless I need to resort to 'paper tape'"
> > pakenned at uiuc.edu                                    --Brynnen Owen 98/9/11
> > 
> > _______________________________________________
> > Cu-wireless mailing list
> > Cu-wireless at lists.groogroo.com
> > http://lists.cu.groogroo.com/cgi-bin/listinfo/cu-wireless
> _______________________________________________
> Cu-wireless mailing list
> Cu-wireless at lists.groogroo.com
> http://lists.cu.groogroo.com/cgi-bin/listinfo/cu-wireless

-- 
David Young                   On the Job Consulting
dyoung at onthejob.net     Urbana, IL * (217) 278-3933