When setting up or growing an e-business, most companies spend a lot of energy and diligence setting up the architecture: acquiring enough servers, hooking up a big enough pipe, choosing the right software technologies, developing the applications, and creating a compelling site design with an intuitive user interface. They spend hours researching and planning to fit these pieces into place. But when it comes to one of the most important decisions of all -- choosing the Internet Service Provider (ISP) that provides that vital link between us and the infrastructure of the Web -- many companies foolishly make the selection based on price alone. Because so many ISPs do position themselves based on price, it's easy to fall into that trap. However, there are several strategic factors essential to consider, including redundancy, reliability, scalability, and support. Because connectivity to the Web is the "spinal cord" of all things Web, choosing a good ISP can be the difference between e-business success and failure.

Know what you're comparing

The normal selection process of an ISP includes looking at a variety of very similar documents: network topography (both owned and partnered), data center maps, rack specifications, support models, and pricing grids. But don't let the paper deluge lull you into thinking you're comparing apples to apples. Although the end product is the same -- getting your site onto the information superhighway -- the roads your ISP takes to get there can be very different.

There are several tiers of service providers. The first tier of providers own their own infrastructure. These are companies such as Exodus, AboveNet, and Level3. A second tier provider is often a customer of the first tier providers and usually offers space and bandwidth only -- no additional services, such as monitoring, managed environments, or professional services. There is also the managed service provider (MSP) and/or application service provider (ASP), such as Loudcloud, Intira, and Resource Peak, which are smaller xSPs that rent space and bandwidth from the first tier. These providers usually offer additional services like Domino Hosting. Check out http://Advisor.com/article/chenb04 for more information on MSPs and ASPs.

The further down the chain you go, the price is generally less expensive. But be sure you are aware what services have been cut in exchange for the price savings. Some of these tradeoffs may be fine for your e-business; other features are mission-critical and non-negotiable for your business. Ask yourself if any of the following services are important to your business, then make sure your ISP provides those features:

• The data center and the network-operating center (NOC) is staffed 24x7x365. This way, you will at least be notified if something goes wrong.

• The staff on hand has the expertise in the particular areas your site needs, such as a networking or database administrator, Web traffic analyzer, NT or UNIX administrator, and is able to immediately handle issues such as rebooting a machine or other things you may not be able to do yourself remotely.

• Backup power and alternate sources of power in case the utility company can't provide power for an extended period of time. Remember, the Web is global, so your customers may not be aware there is a power outage in your area and will still try to access your site.

• Redundant equipment such as routers, switches, power supplies, and Web servers, so there's not a single point of failure. This may significantly add to your monthly charges, but balance this against what would happen if your site went down (and for how long it would be down) if a piece of equipment breaks and has to be ordered.

• High capacity, low utilization. To take into account spikes in usage, you want your provider to have a high amount of available capacity, but a low amount of utilization. In other words, you do not want a provider who, on any given day is using 75 percent of its ultimate capacity -- because one busy day on the Web could bring all its customers crashing down.

• Conservative loading of the electric network. If you've ever plugged a space heater in the same socket as a lamp, you know it's easy to overload a circuit. Ask how the ISP balances the ratio of equipment to power.

• A service level agreement (SLA) based on quantified standards such as network thresholds, response times, and specific remedies if the SLA is not met.

• Not manning the network operations center (NOC), data center, and support center 24x7x365.

• No generators or backup power.

• Lack of failover equipment and hot spares for equipment.

• Liberal down-time policies.

• Undefined or extended response times for problem resolution.

• Using desktop boxes rather than servers for mission-critical operations. Additionally, look for systems manufactured by a reputable company.

• Non-redundant or offbrand networking equipment, including firewalls.

• Overutilization of shared network pipes. Anything above 50 percent capacity should make you look twice.

• Excessive "hops" to the Internet backbone. For example: The ISP may have an OC12 pipe, but it may be connected to another ISP with a T3 ring and then out to the backbone, which will be less capacity/slower than having the pipe connected directly (or short hop) to the backbone.