Fragapalooza sets up an instant network

Take 60,000 square feet of convention show floor, 1,000 PC gamers in multiplayer networked tournaments and $65,000 in prizes, and you have the 12th annual Fragapalooza, Canada’s biggest gaming event and LAN party, held last week in Edmonton.

You also have a unique network challenge.

“At its highest point, you could see 600, 700, 800 people battling it out at the same time,” said Corey Copping, product marketing manager for Hewlett-Packard Canada Co., whose ProCurve network equipment division built the network for the event at Northlands Sportex in Edmonton.

The event pits hundreds of gamers at a time against each other in massive tournaments of games like Call of Duty and Unreal Tournament. Those users need gigabit access simultaneously to the LAN. “There can be no lag,” said Copping.

A standard enterprise network would have to have 4,000 to 5,000 nodes to produce comparable traffic, according to Harley Waterson, ProCurve Networking sales specialist for Western Canada.

And ProCurve techs had only two days to put the network together. “Really, the importance for this type of event is ease of use, ease of deployment and the ability to be put up quickly,” Copping said.

The network architecture ProCurve chose was what Waterson called a flat network: no routing switches between the edge and the core. “You need that quality of service,” he said – the data can’t be held up traversing a router.

There’s an added advantage of simplicity. “(The techs) probably spent more time opening boxes and plugging in” than configuring the network, Waterson said.

It also helped that there were no walls to deal with.

The switch mesh at the core of the network consisted of two ProCurve 3542YLs – one connected to the servers, (including a Quad Xeon E7340 with 16 2.4 GHz CPU cores and 16GB of RAM, running VMWare ESX 3.5, according to the Fragapalooza Web site), one connected to the administration PCs. A ProCurve 3400CL connected to 26 ProCurve 2650 edge switches, offering more than 1,000 ports. The core switches were interconnected with CX4 10 Gigabit Ethernet modules. Shaw cable provided a 1 Gbps connection to the Internet through a firewall connected to the 3542YLs.

Waterson said he might use a similar topology in an enterprise context for a workgroup, but no more than 255 nodes. “The broadband chatter that goes on all the time is always in the background,” he said.

But Info-Tech analyst Mark Tauschek wonders if the distribution layer is redundant in many corporate networks, especially for mid-sized companies.

The three-layer architecture was spawned by Cisco Systems back in the 1990s, Tauschek said. “At the time, it was necessary,” with the network edge populated by hubs and dumb Layer 2 switches. But with more intelligent Layer 2/3 switches at the edge, that routing and processing can be handled there.

“Most small and midsize organizations don’t have that layer,” he said. It’s often still there in larger shops, especially those with Cisco-trained and certified networking staff. The volume of legacy hardware at a larger organization also makes it more likely to have a persisting three-layer architecture, “but it’s not just legacy equipment, it’s also legacy thinking,” Tauschek said.

Related Download
3 reasons why Hyperconverged is the cost-efficient, simplified infrastructure for the modern data center Sponsor: Lenovo
3 reasons why Hyperconverged is the cost-efficient, simplified infrastructure for the modern data center
Find out how Hyperconverged systems can help you meet the challenges of the modern IT department. Click here to find out more.
Register Now