hosting :: sign up > servers > terms

Servers

Servers are Intel Pentium IIIs (dual 1 GHz with 2 GB RAM) with high-speed SCSI or IDE drives. The operating system is Red Hat Linux running Apache Web Server.

Network

Network Operations is managed out of Denver, Colorado. Space is leased from the new State of the Art Verio Network Operations Center (NOC) and their bandwidth, power and connectivity are utilized.

About the NOC:

 Redundant uninterruptible power supplies
 Multiple redundant back-up generators
 Fire suppression
 Raised floors
 HVAC
 Separate cooling zones
 Humidity control
 Multiple levels of security

Connectivity is via 2 OC-12s with Qwest and MCI (soon to be joined by 2 OC-48s) connecting to Verio Chicago and Verio Palo Alto. These are on 4 OC-192 connections, which makes upgrading as simple as adding hardware that is on site - when the need for more bandwidth arises there will be no extended wait while new lines are installed.

Customers connect to 4 Cisco 6509 routers. These routers have multiple connections to our backbone routers. Backbone routers are Junipers with multiple connections to our backbone OCns. The backbone is a Tier 1 backbone. Here is a link to the network map.

The data center has connections to many different Internet backbones including UUNet, Sprint, Cable and Wireless, CRL, Qwest, Exodus, Agis and Net Axs. We also have private and direct peering DS3s set up between our location and that of America Online and PSI-Net. The data center also operates its own DS3 to Mae East to peer with many of the smaller Tier One providers as well as operating another DS3 to the ATM switch located there.

Utilizing multiple Internet backbones results in network connections that are exceedingly dependable. If one of the backbone connections fails, traffic can simply be re-routed over other paths, thereby ensuring the integrity of the network.

In addition, the network runs Border Gate Protocol (BGP4). BGP is used by providers with more than one access point to the Internet. It helps create a truly redundant network. In the event of a lease line failure, the BGP routing session on the bad line will be terminated and traffic will be redirected to a router on a working circuit.

Internal Connectivity

The two main sources of latency for a full-time Internet connection are the user's local area network and the Internet provider's local area network. Our local network is anchored by Cisco 5500 Series ether switches and high-end Cisco routers. This top-of-the-line network hardware ensures that data requests get to their destination and back out of the network as fast as possible. We use ether switches instead of hubs because of their speed and their security capabilities. Only one computer plugged into a hub can talk at a time, whereas all computers plugged into a switch can talk simulatneously. This means more data can travel through a switch than a hub at any given time, and that each server acts as its own node on the network. When each server is its own node on the network it is difficult for hackers to trace data packets with sensitive information (i.e. passwords) to a particular server.

Servers on the network do not share a single path (T3). Instead, the servers are connected to a high-speed ethernet switch. This switch is connected to the core router at the data center. From the core router data is sent back to the end user across the fastest available path. Statically routing traffic over one path creates a single point of failure, whereas our distributed architecture ensures that users can access data extremely quickly and have multiple paths into and out of our network.