Our System & Network

Faster Than OC3 Connectivity

Great support and great features will get you nowhere without a wide pipe. We use to advertise an OC3; now we've gone one better.

Site Host 4 U's Network Operations Center in Baltimore, Maryland is "OnNet" with Frontier Global Center (FGC), which means that we have a direct fiber optic connection between our Cisco 7200 router and theirs. Being OnNet with a Tier-1 provider means that we don't link to a backbone, we are actually on a backbone. We have no phone circuit, and do not use a Telecom link to get to the Internet; instead, we have an in-house connection directly to FGC's ATM fiber node, located a few floors below our servers in the same building. This fiber optic line can handle the bandwidth of a T3 or an OC3, and with FGC's Dense Wave Division Multiplexing (DWDM) technology, it can handle several times the bandwidth of an OC3.

Multiple Backbones:

We share the digital distribution architecture of FGC, which is comprised of more than 25 high-speed private peering connections to major Internet carriers such as MCI, Sprint, UUNET, AT&T, AOL, Best, Erols, and others. FGC also has high-speed links to 8 public exchanges including both MAE East and West and several NAPS. To use an analogy, the private peering connections allow data to travel from New York to LA on a non-stop flight, while the public exchanges require a stop over.

Sometimes the Net is slow...

What happens when your pipe is hooked up to a faucet that just trickles? Sometimes even though your ISP and your web host are both functioning properly, you may still have a slow data transfer rate. The Internet sends information all over the country and the world, through a dozen or more computers on its way to you -- and something is always getting serviced somewhere in that long chain. Here is what we have done to speed things up:

Route Optimization:

We have a large investment in BGP (Border Gate Protocol) technology, which allows the traffic to your site to travel more efficiently by finding the best route for data to travel. On a typical server the traffic always takes the same route from client to server. For them, if there is a bad node, traffic does not get through at all. Because we use BGP protocol, different and more efficient routes are taken between client and server depending on traffic loads and broken nodes. This means our servers automatically look for the fastest route available.

Low Latency/High Throughput:

Often providers operate their networks at three to four times responsible capacity, and as a result the corresponding transfer times reach over 300ms for each hop along the net. Site Host 4 U's network daily average is 6.5% of its capacity, with mid-day peak spikes reaching only 15.5% capacity. Our transfer times range from 15 to 80ms routinely.

HARDWARE & SYSTEM INFORMATION

CPU	Ethernet Connection	Web Server	Operating System
Pentium Pro	100 Base T	Apache Version 1.3.4	Linux Redhat Version 5.2 (Hurricane), Kernel 2.2.0