LINK TO LEARN
PENNSYLVANIA NETWORK
TELECOM 2073
Authors: Diane Gruseck, Louis Guzik, Stephen Sirulnick, Monika Schwartz, Joe Yee
Selection of nodes.
The state of Pennsylvania has already been divided strategically into 29 Intermediate Units (IUs), which each comprise many school districts and cover one or more counties. These divisions represented a natural starting point in selecting nodes. We selected our nodes such that each IU would be represented by a network hub. Each school district within an IU will lease local lines to the IU’s hub, and each school will have dial-up access to its school district.
To choose the town/city that would represent its IU, we looked at Pennsylvania’s state-funded college and universities, reasoning that a major college or university in the area with existing telecommunications resources could allow a backup connection to the network and improve network eliability. Thus, each hub city in our network is home to a state-funded college or university.
Development of topology.
Our first step in designing the network topology was to develop the minimum spanning tree. Because the network is a WAN, we calculated the cost of leased lines. As the crow flies distances between the 29 hubs (in Appendix) were calculated with the use of a Distance Calculator located at http://www.indo.com/distance. Next, we calculated the costs of leasing type 3002 voice grade lines for those distances. Although the final design uses T1 lines rather than voice grade lines, the calculated costs represent a good estimate of the differences in costs of connecting hubs together. The costs can be found in the Appendix. Finally, we used the minimum spanning tree algorithm to generate a minimum spanning tree for our 29 hubs.
After developing the minimum spanning tree, we added extra links to improve reliability and capacity, as described in the following sections.
Capacity Justification
The traffic generated at each hub was estimated using census information taken from the US Census web site (http://www.census.gov/population/estimates/county/). The population for each hub was calculated by summing the populations of those counties serviced by the hub. If a county is serviced by multiple hubs, its population data was divided by the number of hubs by which it is serviced and that value was added to the hub’s total. The hub with the smallest population was then assigned a ratio of 1.0 and all other hubs were assigned a multiplicative ratio factor (in relation to the hub with the smallest population). The results are on the following page.
The assumption for necessary capacity was 1 T1 line for the ratio value 1.0. The capacities of each link were calculated by picking Harrisburg as the central node and totalling the traffic of each hub en route to Harrisburg. If a hub has two paths out to Harrisburg, the traffic was divided in half. Ex. Edinboro - 7.8 (8), Clarion - 2.88 (3), Slippery Rock - 8.96(9), makes each link out of Slippery Rock to have 8.96(9) T1 links. The number of T1 links required for each link and the associated costs are contained in the appendix.
The three major cities in Pennsylvania; Harrisburg, Philadelphia, and Pittsburgh, will be the only cities used as the major node sites for the network. According to recent population data, these three sites have the largest population areas, and they also have the greatest amount of public and private network resources available within a very small geographic area. With these three sites having this amount of population and resources we will now assume that these nodes will be virtually up and running all the time.
We now create a MESH network, using the three major node sites, Harrisburg, Philadelphia, and Pittsburgh, which offers high reliability and good distribution of traffic flow for the entire network. To create the MESH network only two more additional links are needed for it to exist, from Pittsburgh to Harrisburg and from Pittsburgh to Philadelphia.
The Network.
It can be seen, from viewing the attached network diagram, that the general shape of the Minimum Spanning Tree is in a "U" shape through Pennsylvania. The greatest concentration of links, follow a mainly horizontal line across the lower portion of the state. The links from Pittsburgh to Slippery Rock and from Philadelphia to Scranton form the outer sides of the "U" shape. This shape is what we will now try to reinforce.
With the addition of two more links to the Minimum Spanning Tree, from Pittsburgh to Slippery Rock and from Philadelphia to Scranton, even more network reliability and less traffic congestion are possible. If a single link were to go down, and using our assumption that the three major sites will always be up and running, every node in the system is not more than two hops away from a major node site. That major node site is part of the high reliability MESH network. Also this configuration will allow for expansion without too much overlapping of links, which helps to reduce cost. This now produces the final network design that will be used.
Cost Justification for Network
After much consideration to determine the type of line service for this network, T1 line service was chosen over ISDN for several reasons. Cost of installation was the primary consideration, type of service needed was the other concern. According to Bell Atlantic, the cost to install a T1 line was initially $540.00 with a basic rate of $75.00 per month plus $25.00 for every mile from source to destination. As can be seen from the ISDN rates, $700.00 was the installation cost plus there is a rate/ time basis depending on the type of B channel selected for service. Calculations for a cost comparison is unecessary between the two types of service as ISDN is a dial-up service. Due to the nature of the service being provided by the Link to Learn Project, it is most inefficient and ineffective to have a hub interface with another hub in this manner. T1 service would be the best consideration for this as a connection can always be established via this service.
The next step was to determine the number of links that would be needed via each hub. Using average population capacities in each area, we estimated the number of links needed at each hub going to its next node link in the minimum spanning tree. Using costs for T1 service, we have calculated the amount of money needed to initiate the network for the first year, which will be the most costly year in its existance. Costs will decrease as there are no other charges that will be added other than the flat rate of service. Please see attached spreadsheets for data concerning these issues.
The End.