During the course of tomorrow (Wednesday 30th) afternoon we will be completely changing the way traffic is routed between various subnets and buildings on campus. Changes may be made from about 3 PM onwards, but the majority of the work will be done later in the afternoon. All the necessary work should be completed before 7 PM.

If all goes according to plan, most users on campus won't notice the changes. It is possible, however, that all users of the University's network may experience short (i.e. less than ten minutes), intermittent disruptions in connectivity during the course of the afternoon. It is also possible that, at times, connectivity between different subnets or buildings will be slower than normal.

One possible side effect of this maintence work is that user's computers may fail to renew their DHCP leases or browser proxy settings if an outage occurs. If this is the case with your computer, you'll be unable to connect to the network even after networking is completely restored. If you find that you still can't connect to, for example, the University's web server (http://www.ru.ac.za/) after an extended period (half an hour or more), your first step should be to simply reboot your computer.

For those who are interested, details of the work that's going to be happening follows:

This change represents a major shift in the way networking at Rhodes works. It involves migrating the University's routing from a simple static-route based system (as show in http://systems.ru.ac.za/networks/routing.png) to a more complex dynamic system based on the iBGP resdistribution of routes between various routing devices (as show in http://systems.ru.ac.za/networks/bgp-fullmesh.png).

One of the desirable side effects of this change is increased redundancy on the University's network core. Devices connected directly to the core should automatically be able to route traffic appropriately, even in the event of router failures. Coupled with developments that are happening as a result of new African Media Matrix building, the end product should be a core network that's signifcantly more reslient to failure. We still have a number of other changes that need to take place first, however (for instance, the implementation of an inter-switch trunk and VRRP between Struben and the AMM as shown in http://systems.ru.ac.za/networks/smlt.png).

This change will also make management of IP routing on campus significantly easier -- there are currently in the region of a hundred different subnets operating on the University campus and routing between them is becoming increasingly complicated. A complete overview of (proposed) routing on the University's core can be obtained from our newly commissioned iBGP looking glass. This publically accessible looking glass can be accessed at telnet://routeserver.core.ru.ac.za.