As a startup ISP without massive financial backing, we can expect that we will only be able to handle customers with up to 33.6 Kbps modems in the initial months. As outlined below, we are planning on starting with two T1 lines as our access to the internet. Two T1 lines will support 90 33.6 Kbps modems at peak output. By multiplying that 90 peak output number by 10 (the industry standard is a 10:1 total customers to modem ratio), we can reasonably support 900 customers with our two T1’s. Since we will need to support 90 users at peak output, we will need to eventually lease 90 analog phone lines from Southwestern Bell with "roll-over" capabilities. Rolling-over the phone numbers will allow our customers to set their modem software to dial the same phone number with each access. It does not make financial sense to originally lease those 90 lines. We will start by leasing 30 lines, and gradually build up to a target of 90 as our customer base warrants. 

As we grow, we will upgrade our phone lines to "local" T1’s, that can be multiplexed to allow us to handle 56 Kbps modem technology, as well as handle our 33.6 customers. We will not do this initially because of the substantial outlay of capital for a T1 channel bank, as well as the cards required to handle the traffic, altogether costing close to $30,000. After the initial phase of operation, we will also offer our customers the option of connecting to us via ISDN. If ADSL technology emerges as expected, we will add the necessary hardware to support that standard as well.  

As a startup ISP, we can assume that we will need to connect to a backbone with a reasonably large bandwidth. Our best choice in the initial phase is to lease T1 lines from a large telecommunications company such as Sprint. Sprint's backbone is 100% fiber optic, has redundant SONET technology that helps reroute IP packets in the case of network outages, and we feel we will get better customer service from Sprint than from a larger company such as AT&T.  This will allow us to connect to Sprint’s backbone with a reasonably good capacity of 3.088 Mbps, allowing us to handle 90 customers at peak output, and have a customer base of roughly 900. Charging $19.95 per customer will create revenues of close to $18,000 per month. 

T1 lines are not cheap. Sprint will charge about $1,000 each to install the lines and then $1,200 per month to lease those lines. So in the first month, our bill will be about $4,400, and then about $2,400 thereafter. 

As we approach the threshold of 900 customers, we will have to make some decisions. In order to support more customers, do we want to add another T1, or possibly upgrade to a fractional T3, or even a whole T3? A T3 will give us a bandwidth of 45 Mbps, but we will have to have a large increase in customers to justify such an expense. A T3 would have the capacity to handle over 13,000 customers if they were all using 33.6 modems, which would not be the case. But with a T3, we could attract customers who are looking for an ISDN connection of up to 128 Kbps, and we could even attract customers looking to connect to the internet through us using a T1 line. A T3 would cost over $5,000 to install, and over $10,000 per month to lease.  We will need a substantial customer base increase to justify a T3. We will probably go with a half of a T3 in the second stage of our company’s existence, and possibly upgrade to a full T3 later as our customer base warrants. 

This is one of the easier decisions we will have from a hardware/technology standpoint. Our hardware specialist has come to know and love the ease and benefits of 10BaseT Ethernet. For roughly $3,500, our company can install 10BaseT with 16 ports, what we will need initially. This price would include the hub, stack, and NIC cards. The internal network would be 3 times as fast as our backbone connection, so we would not have to worry about network traffic problems.

There is an old saying that goes something like this: "There are only three things to consider when you start a business: location, location and location." While the typical business needs to be wary of its location for the usual reasons such as proximity to customers, whether customers can see their business from the roadway, or whether the business is located in a good neighborhood, our ISP will be concerned with location for different reasons.

There are two schools of thought on how to run a "mom and pop" style ISP, like we will be starting. The first theory is to have the conventional rented industrial space, where all of the functions of the ISP including hardware maintenance, billing, sales, and other everyday activities are handled from one central location. All of the equipment would be located there, as well as the administrative offices, and we would have a T1 connected from this site. For the purposes of this project, this is the approach we will incorporate.

The other theory is to have all of the hardware, including the modem block, core servers, RAS server, Ethernet, CSU/DSU, and such located in a room in a telco Point of Presence (POP), where other ISP's might also locate there equipment. In this room there is a direct connection to a T3, and we would lease our T1 from this company. We would also have to rent some office space at another location, and we could run all of the equipment located at the POP remotely from our central office, via telnet and web browser interface. This would require us to occasionally, maybe even frequently, go out to this POP to fix equipment, and deal with other problems, but would virtually eliminate such problems such as thunderstorm interference that can effect telephone wires.

The second approach could be more cost effective, but I feel better about having our equipment and offices at one location. Expansion could be easier, we only have to worry about security problems for one building space, and hardware problems could be dealt with much faster.