Our architecture is a complete Cisco Network. We utilize fully redundant Cisco 6509 routers with the latest SupII/MFSCII cards with 30 mm pps throughput (redundant routers and redundant cards within the router) at the core and state of the art 3550 L3 switch / routers as aggregation routers. To further maximize redundancy we utilize distributed 2924 XL EN switches to ensure that single chasis failure will have minimum downtime effects on our servers - unlike many of our competition who aggregate hundreds of servers in a single chasis based switch like the 6500 or 5500 which if it fails its cpu or power - it takes down many customers This approach gives us maximum power failure protection and switch failure protection to distribute the load and risk.

We also utilize VLANS to prevent customer IP space hijacking and conflicts. By keeping our network as flat as possible - we minimize route convergence and latency in the backbone and speed the packets to the best provider through BGP4 on the core.

Network - we believe in a few large pipes of quality providers to handle spikes in traffic and the occasional DoS attack as well as unknown traffic patterns in the case of a primary link failure. This is why our network minimum Internet backbone connection is 1 gig pipes. We currently have 5 backbone providers and are publicly peered with 12 providers including earthlink. Our current network consists of Gigabit links to Abovenet, XO, PCC-BTN, Telia, Level 3 and the Atlanta Internet Exchange public peering point. Your customers and you will not get bottlenecks due to small pipes. By having less providers with a larger burst capacity we are able to handle DoS attacks better than many of our competitors that have chosen many smaller pipes. This is because a DoS attack will many times concentrate itself on one inbound or outbound connection and can swamp it if it is only a 100 meg or an oc3. We do not provision less than GIG circuits on the backbone. It also allows us plenty of room to fail back and forth from provider to provider without regards to where the traffic is going to prefer since any one of our links has far sufficient size to handle all of our traffic.

We actively monitor our routers for key factors including temperature, cpu and memory utilization and monitor our connections for transit latency and uptime. We are notified immediately in the event of a process going out of tolerance or an outage through our automated systems which allows us to respond quickly with corrective action.

We also monitor all our servers in the data center for uptime and immediately notified in the event of an outage on the server.