Once your network starts to get really big, or you start to consider 'the internet' as your network, you need tools which dynamically route your data. Sites are often connected to each other with multiple links, and more are popping up all the time.
The Internet has mostly standardized on OSPF (RFC 2328) and BGP4 (RFC 1771). Linux supports both, by way of gated and zebra.
While currently not within the scope of this document, we would like to point you to the definitive works:
Overview:
Cisco Systems Designing large-scale IP Internetworks
For OSPF:
Moy, John T. "OSPF. The anatomy of an Internet routing protocol" Addison Wesley. Reading, MA. 1998.
Halabi has also written a good guide to OSPF routing design, but this appears to have been dropped from the Cisco web site.
For BGP:
Halabi, Bassam "Internet routing architectures" Cisco Press (New Riders Publishing). Indianapolis, IN. 1997.
also
Cisco Systems
Using the Border Gateway Protocol for interdomain routing
Although the examples are Cisco-specific, they are remarkably similar to the configuration language in Zebra :-)
Please, let me know if any of the following information is not accurate or if you have any suggestions. Zebra is a great dynamic routing software written by Kunihiro Ishiguro, Toshiaki Takada and Yasuhiro Ohara. With Zebra, setting up OSPF is fast an simple, but in practice there's a lot of parameters to tune if you have very specific needs. OSPF stands for Open Shortest Path First, and some of its principal features are:
Networks are grouped by areas, which are interconnected by a backbone area which will be designated as area 0. All traffic goes through area 0, and all the routers in area 0 have routing information about all the other areas.
Routes are propagated very fast, compared with RIP, for example.
Uses multicasting instead of broadcasting, so it doesn't flood other hosts with routing information that may not be of interest for them, thus reducing network overhead. Also, Internal Routers (those which only have interfaces in one area) don't have routing information about other areas. Routers with interfaces in more than one area are called Area Border Routers, and hold topological information about the areas they are connected to.
OSPF is based on Dijkstra's Shortest Path First algorithm, which is expensive compared to other routing algorithms. But really is not that bad, since the Shortest Path is only calculated for each area, also for small to medium sized networks this won't be an issue, and you won't even notice.
OSPF counts with the special characteristics of networks and interfaces, such as bandwith, link failures, and monetary cost.
OSPF is an open protocol, and Zebra is GPL software, which has obvious advantages over propietary software and protocols.
Compiled with CONFIG_NETLINK_DEV and CONFIG_IP_MULTICAST (I am not sure if anything more is also needed).
Get it with your favorite package manager or from http://www.zebra.org.
Let's take this network as an example:
----------------------------------------------------
| 192.168.0.0/24 |
| |
| Area 0 100BaseTX Switched |
| Backbone Ethernet |
----------------------------------------------------
| | | |
| | | |
|eth1 |eth1 |eth0 |
|100BaseTX |100BaseTX |100BaseTX |100BaseTX
|.1 |.2 |.253 |
--------- ------------ ----------- ----------------
|R Omega| |R Atlantis| |R Legolas| |R Frodo |
--------- ------------ ----------- ----------------
|eth0 |eth0 | | |
| | | | |
|2MbDSL/ATM |100BaseTX |10BaseT |10BaseT |10BaseT
------------ ------------------------------------ -------------------------------
| Internet | | 172.17.0.0/16 Area 1 | | 192.168.1.0/24 wlan Area 2|
------------ | Student network (dorm) | | barcelonawireless |
------------------------------------ -------------------------------
Don't be afraid by this diagram, zebra does most of the work
automatically, so it won't take any work to put all the routes up with
zebra. It would be painful to maintain all those routes by hand in a day
to day basis. The most important thing you must make clear, is the network
topology. And take special care with Area 0, since it's the most important.
First configure zebra, editing zebra.conf and adapt it to your needs:
hostname omega password xxx enable password xxx ! ! Interface's description. ! !interface lo ! description test of desc. ! interface eth1 multicast ! ! Static default route ! ip route 0.0.0.0/0 212.170.21.129 ! log file /var/log/zebra/zebra.logIn Debian, I will also have to edit /etc/zebra/daemons so they start at boot:
zebra=yes ospfd=yesNow we have to edit ospfd.conf if you are still running IPV4 or ospf6d.conf if you run IPV6. My ospfd.conf looks like:
hostname omega password xxx enable password xxx ! router ospf network 192.168.0.0/24 area 0 network 172.17.0.0/16 area 1 ! ! log stdout log file /var/log/zebra/ospfd.logHere we instruct ospf about our network topology.
Now, we have to start Zebra; either by hand by typing "zebra -d" or with some script like "/etc/init.d/zebra start". Then carefully watching the ospdfd logs we should see something like:
2002/12/13 22:46:24 OSPF: interface 192.168.0.1 join AllSPFRouters Multicast group. 2002/12/13 22:46:34 OSPF: SMUX_CLOSE with reason: 5 2002/12/13 22:46:44 OSPF: SMUX_CLOSE with reason: 5 2002/12/13 22:46:54 OSPF: SMUX_CLOSE with reason: 5 2002/12/13 22:47:04 OSPF: SMUX_CLOSE with reason: 5 2002/12/13 22:47:04 OSPF: DR-Election[1st]: Backup 192.168.0.1 2002/12/13 22:47:04 OSPF: DR-Election[1st]: DR 192.168.0.1 2002/12/13 22:47:04 OSPF: DR-Election[2nd]: Backup 0.0.0.0 2002/12/13 22:47:04 OSPF: DR-Election[2nd]: DR 192.168.0.1 2002/12/13 22:47:04 OSPF: interface 192.168.0.1 join AllDRouters Multicast group. 2002/12/13 22:47:06 OSPF: DR-Election[1st]: Backup 192.168.0.2 2002/12/13 22:47:06 OSPF: DR-Election[1st]: DR 192.168.0.1 2002/12/13 22:47:06 OSPF: Packet[DD]: Negotiation done (Slave). 2002/12/13 22:47:06 OSPF: nsm_change_status(): scheduling new router-LSA origination 2002/12/13 22:47:11 OSPF: ospf_intra_add_router: StartIgnore the SMUX_CLOSE message by now, since it's about SNMP. We can see that 192.168.0.1 is the Designated Router and 192.168.0.2 is the Backup Designated Router
We can also interact with the zebra or the ospfd interface by executing:
$ telnet localhost zebra $ telnet localhost ospfdLet's see how to view if the routes are propagating, log into zebra and type:
root@atlantis:~# telnet localhost zebra
Trying 127.0.0.1...
Connected to atlantis.
Escape character is '^]'.
Hello, this is zebra (version 0.92a).
Copyright 1996-2001 Kunihiro Ishiguro.
User Access Verification
Password:
atlantis> show ip route
Codes: K - kernel route, C - connected, S - static, R - RIP, O - OSPF,
B - BGP, > - selected route, * - FIB route
K>* 0.0.0.0/0 via 192.168.0.1, eth1
C>* 127.0.0.0/8 is directly connected, lo
O 172.17.0.0/16 [110/10] is directly connected, eth0, 06:21:53
C>* 172.17.0.0/16 is directly connected, eth0
O 192.168.0.0/24 [110/10] is directly connected, eth1, 06:21:53
C>* 192.168.0.0/24 is directly connected, eth1
atlantis> show ip ospf border-routers
============ OSPF router routing table =============
R 192.168.0.253 [10] area: (0.0.0.0), ABR
via 192.168.0.253, eth1
[10] area: (0.0.0.1), ABR
via 172.17.0.2, eth0
Or with iproute directly:
root@omega:~# ip route 212.170.21.128/26 dev eth0 proto kernel scope link src 212.170.21.172 192.168.0.0/24 dev eth1 proto kernel scope link src 192.168.0.1 172.17.0.0/16 via 192.168.0.2 dev eth1 proto zebra metric 20 default via 212.170.21.129 dev eth0 proto zebra root@omega:~#We can see the zebra routes, that weren't there before. It's really nice to see routes appearing just a few seconds after you start zebra and ospfd. You can check connectivity to other hosts with ping. Zebra routes are automatic, you can just add another router to the network, configure zebra, and voila!
Hint: You can use:
tcpdump -i eth1 ip[9] == 89To capture OSPF packets for analysis. OSPF ip protocol number is 89, and the protocol field is the 9th octet on the ip header.
OSPF has a lot of tunable parameters, specially for large networks. In further ampliations of the howto we will show some methodologies for fine tunning OSPF.