|
|
|
Network Basics
Internet2, for most purposes, differs little from the "commodity Internet." Both networks use the TCP/IP protocol suite and the same type of fiber infrastructure. Commodity Internet applications run over Internet2 and vice versa. The main differences between the two include 1) the amount of bandwidth available (often called "the size of the pipe" as in you can move more water through a big pipe than a small one) and 2) some extra services such as Quality of Service, video conferencing, etc. The bottom line is that those with access to Internet2 can run applications with higher bandwidth demands. What applications fit this bill? The most obvious are digital video applications (such as video on demand and video conferencing), distributed high performance computing (often called grid computing), and large-scale visualization and animation of those computed results.
While there is nothing technical that prohibits these sorts of applications from running on the commodity Internet, they may not run very well. This is generally due to excessive Internet traffic congestion, odd routing patterns caused by congestion and boundaries or differences between ISPs, and bottlenecks caused by one or more low-bandwidth arteries within the network route path. (Some have noted that on a typical afternoon, packet delays on the commodity Internet can easily reach several hundred milliseconds and effective bandwidth can dwindle to 5 Kbps or less). So Internet2 applications are actually those that are made practical by the infrastructure (both fiber and people) known as Internet2.
The following image shows all of the Internet2 Gigabit Points of Presence (GigaPops.) (You can click on the image to download a large pdf version.) The GigaPops are routing points with high bandwidth pipes between them.
The Internet2 bandwidth, between its GigaPops, is OC-12 (622 Mbps.) The connection between Internet2 and our gateway to international research networks, STARTAP, is OC-48 (2.488 Gbps.) Aggregation points further distribute Internet2 to state, regional, or research networks and are generally OC-12 or OC-3 (155 Mbps.) A detailed graphic of the Internet2 Abilene Network can be found on the Internet2 Abilene Network Operation Center site.
So what do you do if you can't seem to make a connection or if your connection seems to be working poorly? There are two likely causes of such things: 1) the unit at the other end is down or 2) some segment of the network is having trouble. Two very simple tools are available to help figure this out.
The first tool goes by the general name of ping. Ping is useful when you are unable to get a system to respond after you've tried to reach it. Doing a web search for ping tools will likely return many responses about things that provide the basic ping service: to see if the host at a particular ip address is alive and working. Most workstations and PCs have the basic ping program available. For most Windows systems you can simply open a command prompt window and type "ping <destination ipname or ipnumber>" where destination ipname or ipnumber is the ip name or number of the system to which you are trying to connect. Ping, in its simplest form, will just respond with an "ipname or ipnumber is alive". More fully configured versions will allow you to define how often to attempt the ping, how large a packet to use with the ping, and how many times to send the packet. A round-trip time is returned for each packet sent. GUI versions are available that will store these results for statistical use and even provide plots of those statistics.
If ping does not reply with a positive acknowledgement, it can mean several things. You may have used the wrong ip name or number. Its always wise to double check that first. If you're sure of the ip name or number, a ping failure could mean that the system is down. But it could also mean that some part of the network is down or is experiencing problems. The tool we want to look at for checking the network is called traceroute.
Traceroute is similar to ping in that basic versions are found on most systems and full-featured versions are available at a small cost. Again, from your command prompt window, simply type in "tracert <ipname or ipnumber>". You will receive a list of "hops" along with the millisecond delay associated with each hop. Each hop is typically a router. The combination of hops from your system to the destination system is your path. Delays at each hop, for Internet2, will range from about 10-40 ms. Delays in the 100+ ms range become noticeable for video conferencing; those in the 1,000 ms range are debilitating.
One last thing to note is that the paths to and from a destination are not the same. That is, the route from A to B is not the same as the route from B to A. So, while one end of a video conference may not be seeing any difficulties, the other might. Therefore when possible, it is a good idea to have both ends traceroute to the other and then compare notes.
These tools are very useful in helping to isolate problems. With the results, you may be able to fix them yourself. If you can't, its time to contact your network engineer and provide them with the information you have collected. It will get them off to a good start.