World Youth Day, 2008
WYD08 was a huge project! One of our partners, eIntellego, was awarded the contract to supply Internet services to the Media Tribunes at Randwick and Barangaroo, and to the International Media Center at the Sydney Convention and Exhibition Centre. Rendrag Networks was contracted to assist with the delivery network design, and building a web-based management tool to manage all delivery devices within the IMC and media tribunes.
The core network at each site consisted of dual Gigabit internet links from Telstra Internet, each terminating on a dedicated router, cross-linked to eIntellego-provided routers, running BGP sessions between each. This provided full failover should one link drop out, or a hardware error cause a router to drop out. The delivery network on the eIntellego side was running OSPF to allow for relatively simple network configuration, and allowing static IP ranges to be moved around the network as needed, without a noticeable lag time.
To allow for a 'simple' network model, we implemented a 'pod' model, where each physical area would be served by a 'pod', which had one or more physical interconnections back to the core of the delivery network (for physical-link failover). Each 'pod' would have one or more switches, served by one or more routers. The 'magic' here was that all allocation of switch/router pairing within the pods was managed dynamically by the 'Imperium' web-based management software.
Imperium
Imperium was a postgres/php5 web application, which allowed eIntellego network engineers to build a template for each switch and router. Each template would have a 'static', 'dynamic', 'enabled port' and 'disabled port' section. Once all templates were entered into the system, a 'pod' would be created, and switches and routers added to the 'pod'. IP Addressing and login details for each device are assigned, and then 'ports' are added. A 'port' is defined as a numbered port at a physical location. This allowed for a 'friendly' labelling convention to be crosslinked to switch ports and router-backed vlans.
Every switch port was assigned to a vlan, and set to one or more routers within the pod. Using vlans backed to routers, allowed us to provide each port with a set bandwidth limit (i.e. 256kbps/512kbps/1500kbps/4mbit/etc), and a choice of a whether each port was a standard NATted IP, a public static IP, or assigned to one of the set vlans (WYD Staff, AFP, ASIO, or one of the shared VLANS which a number of the larger media organisations required to easily spread their network acros their offices through the IMC).
Once all ports were defined by the network engineering staff, standard helpdesk staff were able to provide 24x7 support to media staff, able to add/edit/remove port allocations, without the need for any cisco or network engineering expertise. The Imperium software queued changes made by the helpdesk staff, and automatically connected to the pod switches and routers to apply the appropriate IOS and CatOS commands to enable the selected profiles.
In the event of a hardware failure, a full configuration dump could be generated by the software, to be loaded into a clean replacement device, ready for fast deployment into the affected pod.
On the whole, the event was a major sucess, with only one pod failure late at night, which the helpdesk staff were able to rectify within 5 minutes.
 |
|
 |
|
 |
|
|
|
 |
 |
 |
