Network Computingwww.networkcomputing.com

First, realize that there's no one all-encompassing super ISP. Rather than hunt down the most full-featured ISP, find a provider that does a good job of servicing your organization's needs.

Second, do some planning. Your choice of an ISP should be based, at least in part, on your organization's Internet-related goals. Formulate your agenda before you go shopping. For example, what is your organization's goal for Internet connectivity? Will the connection be used for Internet-based services for internal users, or to offer the organization's resources to the Internet community, or both? What specific services will you offer? Will they be housed internally? How mission critical will they be? ISPs can help with general requirements, but they cannot answer the questions that need to be answered by those within your organization.

One Size Doesn't Fit All Wouldn't it be nice if the solution to that leaky-faucet project--on which you spent an entire Saturday working--was as simple as walking to the hardware store and asking for a new pipe? Unfortunately, you can't just plug in a pipe and expect it to work. Similarly, a garden hose won't do the job of a fire hose, and you certainly don't want a fire hydrant as your faucet. The same is true for your "pipe" to the Internet.

Transfer rates range from the slow analog connections of a modem to the rocketing rates of a digital DS3/T3 (45-Mbps) connection and beyond. The costs for these varying speeds range from the relatively inexpensive to the positively pricey. The two most important factors to consider when choosing that initial connection rate are bandwidth needs and upgrade paths.

If, for example, your organization has 1,000 users and the primary use for the Internet connection will be e-mail, then a T1 connection might be overkill. In contrast, an organization with five people and the coolest Web site on the Internet may not be adequately served by several T1s. There is no way to determine your exact needs, but some simple investigating can turn up a reasonable estimate. For example, look at the applications that will be used. File transfers, Web-related transfers and multimedia are the biggest bandwidth hogs; e-mail and gopher-base d services usually aren't as greedy.

Once your site develops devoted followers, available bandwidth may diminish. Upgrade paths are vast, so it's important to look at an ISP's upgrade capabilities and strategies. Many ISPs have seamless upgrade methods that require only a few seconds of downtime. For those wishing to start small--knowing they will eventually grow--a clever, yet amazingly simple method is to lay an initial T1 circuit, but activate only the first few channels. The monthly bill from the local telephone company will be a bit higher, because you're paying for a bigger circuit to the provider, but the ISP will charge you only for the bandwidth activated. Using this model, upgrades from a 56-Kbps connection to the full use of the T1 will require just a simple configuration change on the CSU/DSU--no new lines, routers or ports.

How Much for the Hose and the Hydrant? The actual cost of connecting a site to the Internet--provider-related costs only, not the price for OS upgrades, ser ver patches, aspirin or bar tabs--generally involves three factors: initial start-up costs, regular monthly costs and any costs associated with additional " nonstandard" services. Ask the provider about upfront costs from the get-go.

One-time start-up costs typically include installation of the actual circuit, cost of necessary hardware (usually a CSU/DSU and a router) and other costs the provider tacks on for overhead. It's important to know a provider's hardware policy: Some will let you lease hardware; others will sell it to you. If you plan on using, programming or monitoring the Internet router at your site, make sure the ISP doesn't insist on maintaining managerial control over the equipment. And to be on the safe side, if you're partial to one brand of router, check that it's OK with your ISP.