[WLANware] Selling the idea of OLSR

[Juergen Neumann]

Hi Dan,

> I am giving a presentation to my wireless group, Melbourne Wireless,
this
> Friday on OLSR.  We currently use OSPF and a traditional routed
network
> where each node has multiple radios, each with a separate /28 subnet.
> Most
> of our links are point-to-point or point-to-multipoint (client-to-AP).
A
> lot of people are very sceptical of the idea of using a /16 netmask on
an
> interface and of using Omni-to-Omni links.
> 
> Here are some of the criticisms I'll need to address:
> 
> * Mesh networks are not scalable - they cause too much CPU load on
small
> routers.

Yes, we have experienced that again and again. But so far, the main
developer of the freifunk.firmware - Sven-Ola - has always managed to
solve this issue in time by optimizing the system, before(!) it really
sucked. ;-)

Here are some interesting outputs from my WRT54-GL:

  _______                     ________        __
 (       ).-----.-----.-----.)  )  )  ).----.)  )
 (   -   ))  _  )  -__)     ))  )  )  ))   _))   _)
 (_______))   __)_____)__)__))________))__)  )____)
          )__) F R E I F U N K  F I R M W A R E

root at k9wrt:~# route -n | wc -l
+ exec ip route
    190

-> So with a total of 190 routes my WRT has to calculate the CPU load
is:

root at k9wrt:~# top
Mem: 10560K used, 3820K free, 0K shrd, 888K buff, 4376K cached
Load average: 0.21, 0.29, 0.22    (State: S=sleeping R=running,
W=waiting)

  PID USER     STATUS   RSS  PPID %CPU %MEM COMMAND
  608 root     R        796     1 13.9  5.5 olsrd
__________________________________________________

We made tests with about 500 nodes by connecting several community
networks via VPN tunnels. The result was a CPU load of about 50%. So far
we have always managed to improve our technology before we ran into
serious problems. 

There are already new and very promising ideas to go beyond these
numbers but until now our network in Berlin has "only" about 300 nodes -
which are not all sharing the same meshcloud - and everything workes
fine.

Though it is a german website you might want to check the map on our
website: 

http://olsrexperiment.de/index.php?option=com_wrapper&Itemid=94&lang=

(The list of IPs on the right indicates nodes that haven't registered to
the map yet. The number on top is the number of nodes the mapping-system
is connected to. The map is live!)

So to us it works pretty well so far.

> * Omni-to-Omni links have limited range and the population density in
> * Melbourne is not sufficient for a critical mass to form a useful
mesh
> * Omni antennas pick up too much RF noise for usable links

That is very sweeping! Of course the link quality may always vary
depending on the HF surrounding. Point-to-point HF-links are of course
better most of the times, but in a high-density meshed network you'd
need a lot of accesspoints, antennas and spare time to build that. Many
nodes here have 5-7 neighbours. And yes: some of the links are rotten!
But others are very nice. 

OLSR does nor depend on ad-hoc-mode, neither on the type of antenna (not
even on wireless). Using wireless, ad-hoc and omnis is just the cheapest
and most open meshing approach! So you may always mix the type of
connections just by your local needs. We do also have point-to-point HF
links (some are managed mode, too) to bridge longer distances and on
certain links that have kind of a "backbone" function.

We have a lot of fine links using omnis, too, where the distance and the
line-of-sight and the HF noise is ok - just mix it by your local needs.

> * A network where everyone is using a /16 netmask will be like a giant
LAN
> and everyone will be swamped with everyone else's broadcast traffic

Yes, there is a certain amount of shit on the air. But there's always
pros and cons. Educating the users and configuring PCs and firewalls
helps of course ;-). 

> * Using OLSR on every node is like trying to create a city-wide
network 
> with WDS.

Definitely NOT! See all my points above and get more reports from other
communities using OLSR like eg. Vienna (http://funkfeuer.at) or Leipzig
http://leipzig.freifunk.net and Weimar http://weimar.freifunk.net.


Using the freifunk-setup the way we use it in Berlin is just the most
open approach to get people to join the network by themselves. We have
no centralized administration and those who want can just buy an
acceesspoint, register to get their IP address, flash their router and
if there is a neighbour to reach it's done!

This of course results in some flaky links, some out of date router
configurations, some bad transmission rates, some abusers, and some
other problems. 

But over all - and that is what counts here! - it's the most flexible,
dynamic, sustainable and scalable approach to build a truly
non-centralized community owned network!

The many users of our network here in Berlin can tell ...

Just my two cents,

JuergeN