Routing

New version of OLSRd hitting RC

We're looking at some performance improvements and some new features. Should be packaged up for platforms of interest shortly.

hi all,

Top of tree has been tagged to OLSRD_0_5_6_RC1.

source tarballs can be downloaded at:

http://www.olsr.org/releases/0.5/olsrd-0.5.6-rc1.tar.bz2
MD5 49d55a68d1b2b2ac040f4c4df179ba69

http://www.olsr.org/releases/0.5/olsrd-0.5.6-rc1.tar.gz
MD5 d5161c51b8a3c75a1f19db1ff86c2e67

i am asking now the package maintainers to download, compile and test
on their target system and provide feedback (both positive and negative
feedback is welcome).

tentative release target for 0.5.6 will be May 10th.

/hannes

-- 
Olsr-dev mailing list

brain dump about the networking experience at LCA08

One of the key things about helping at an event like LCA08 is documenting things so that the team won't have to suffer the things you went through!

So begins the process of dumping ones brain to the Wiki...

Hopefully this might help others in their journey into building their networks =)


OLSR 0.5.3 binary package for Ubuntu/Fiesty 7.04

Got busy with a compiler today and built up a package for Ubuntu 7.04 or Fiesty.

ftp://ftp.adelaide.edu.au/pub/users/khawti01/olsrd/ubuntu-feisty/


Managing Collections of Sub-Networks

When a router receives a packet of information (as it does thousands of times per second) it inspects the IP Address of that packet, and based on that, can make decisions on which physical path to send it through.

This process is known as routing the packet. Special software protocols have been created to support the routing of network traffic between subnetworks. These protocols operate with ‘smart’ algorithms that can adapt the flow of network traffic when problems occur. Protocols such as Border Gateway Protocol (BGP) and are widely used across the Internet today.


Managing Networked Computers

Computers on a network communicate with each other using a special numbered addressing system. This is known as Internet Protocol (or IP). Each computer has a unique IP Address, which is four sets of numbers from 0-255, separated by a decimal point. This gives a total possible address space from 0.0.0.0 to 255.255.255.255. For example: 203.152.23.202 is an Internet IP Address.

IP addresses can be grouped into smaller logical segments of the entire address space, which can then be assigned to computers in a LAN. Breaking up IP Address space into smaller sub-networks is known as Subnetting.

This allows for logical control of data between LANs, and maps well onto the physical network design outlined.


Physical Network Structures

Groups of computers can be connected in many different configurations.

A simple Local Area Network (LAN) consists of a few computers, which are directly connected to each other.

This allows for high-speed communication between the participating machines, but offers little flexibility for adding more clients. Networks like this are generally confined to a relatively small area, such as a home, business or office.

As more clients are added to the LAN, we require more efficient control of network traffic. This is achieved through the deployment of devices such as Switches and Routers.


Air-Stream's Routing

Air-Stream primarily uses BGP for routing around the domain, some AP owners choose to run another protocol such as OSPF/RIP locally then redistribute into the BGP network, this is fine.

There are filters run on the Air-Stream operated routers to control routing updates and what can and can’t be routed, it’s generally fairly open.

Each AP is assigned its own Private AS (Autonomous System) and AP owners have full control of their own AS in regards to what goes in and out. Initally we started off with OSPF but ran into multicast and topology issues, BGP allowed us to work around this while providing added control over the network so it was a win win situation.

As previously mentioned we like to summarise addresses where possible at the routers and have a good design in place to allow for huge growth. We are able to provide BGP templates for people to use as configs, it really isn’t that hard and configuration isn’t overly advanced for the nodes, about the most interesting thing is that they run in a route reflector client mode due to the topology and iBGP.