Tom DeReggi wrote:
Trie I did not offer any backup data. But use your immagination. Its
all in one place, easy to check, easy to document, easy to configure,
easy to backup, etc.
What does mesh offer for better complete central management?
You seem to be suggesting that I simply haven't looked for information
to back up your argument. Not sure why that makes sense to you. Anyway,
I am not making arguments based upon information I read somewhere. My
company operates a very large network that makes use of mesh, star, and
ring network architectures. Some of it is fiber-based, while other parts
are wireless. We are a highly technical, but practical company. In other
words, we do a vast amount of research before doing field trials. After
we are satisfied that the technical works in the field the way we expect
and ultimately want, only then do we deploy it. I can make intelligent
statements in regard to mesh because of this. It doesn't appear you have
done nearly the research we have and it doesn't appear you have any
significant mesh deployments. I suggest you field trial the technology
in a meaningful way before dismissing it.
In regard to your actual question, I would request that you be more
specific. We manage all of our network devices centrally using SNMP
regardless if they are mesh or not.
> I think you may be mixing too many arguments.
I may be mixing up typical deployment models using MESH with MESH
Technology.
It also depends on your definition of MESH.
Cisco defines a mesh network as a communications network having two or
more paths to any node. I would agree with that definition. How would
you define mesh?
I admit, I made a generalization of a typical way MESH would be
deployed, in my arguements.
Deployed at street level, so many short hops were required to get
coverage and get around NLOS obstacles, in a dense city environment.
That may be true if the mesh didn't have any dedicated backhauls. We
using P2MP systems to backhaul our mesh, which allows us to limit the
number of hops of any one particular path.
A network that made its own intelligent routing decissions, that may
not always be the most intelligent compared to the human mind's
decissions.
Meshs don't have to make their own routing decisions. You can statically
route a mesh if you want to. I don't think I will agree that a human is
better suited to the job though.
But is that really MESH? Technically you could call any multi-path
routed network, MESH. I call my network a routed network using
triangulation.
But I would not call it MESH. But it very well could be considered
similar to MESH.
Our industry peers use the term mesh in this context, so it appears
quite appropriate.
What criteria does your network OS sue to deterine routing changes?
Measure highest packet loss? measure most amount of available
bandwdith? Measure least amount of average bandwidth? Measure shortest
path? Lowest latency? Lowest cost ($) transit or transport provider
path? And how many can they consider togeather to make the best
overall decission?
I'd be interested in hearing more about what you are doing with MPLS
in your design.
MPLS traffic engineering allows you to use any number of combinations of
criteria. In fact, Cisco sells whole books on this very subject.
Also understand this is a Wireless list, not a fiber list. The design
flaws of MESH over fiber (fast packet-loss less links) is a completely
different animal with different challenges than MESH in Wireless.
I disagree. While there are certainly important differences between
fiber and wireless, network architecture wish the communication medium
is generally less important.
I recognize that MESH is at a new stage of being more than just the
implementation of RIP2. (Allthough early MESH was not much more than
RIP).
Tropos's implementation certainly doesn't fit that description and they
have been around from the early days of wireless mesh.
Thats a very bold statement, that is not true. However, that does not
mean I do not recognize the benefits of the advanced design of MPLS
networks.
What do you mean it isn't true. Of course it is! Name one tier 1 ISP
that doesn't have an MPLS network or is working on having one.
How do you figure? I sure hope the network design that was getting
proposed, was something they would take the time to evaluate, in
making their decissions.
Anyone would look at there assets to locate gear, and consider that
into their design. Thats step 1 of any wireless network design.
We consider mesh for its technical merits and this thread started in
that regard. We have nothing to do with munis and yet we do a good deal
of mesh. It seems very simple that mesh as a technology and one market
segment are two separate issues. Discuss muni issues in a non-technical
thread.
-Matt
--
WISPA Wireless List: wireless@wispa.org
Subscribe/Unsubscribe:
http://lists.wispa.org/mailman/listinfo/wireless
Archives: http://lists.wispa.org/pipermail/wireless/