Prospective shared hosting clients must therefore ensure that they test the technical capacity of any host thoroughly before they procure their services. Advanced testing of a potential host will reveal whether the solutions they provide are reliable enough for your high-traffic site. Remember that your Web host must be trusted to provide solid network infrastructure. If you select a host that cannot provide robust connectivity, then your site's availability to the world will suffer. For this reason, informed consumers will evaluate potential hosting firms before they sign-up.

Testing ensures that consumers will not waste their good money on bad services. Reliable testing results can be obtained through the use of sophisticated network tools that monitor hosting performance. Such tools will determine how often a host's servers experience outages and will generate a list of probable reasons why hosting services are unreachable. It is advantageous for you to use such tools to ensure that the host you select will provide minimum downtime. Most hosting firms boast about their relentless commitment to excellent service and server responsiveness, and usually the crowning jewel of this commitment is 99 per cent uptime.

But while most hosting operations use this promise of incredible uptime as a hard sell, few consumers actually test whether these pledges are true. Smart consumers of hosting services, on the other hand, are the first to authenticate these service guarantees. They usually consult the services of an established server monitoring services such as NetMechanic, provided by Keynote Systems.

NetMechanic provides an integrated suite of tools that detect problems with your Web site. The company's "Server Check Pro" product is an excellent choice for ensuring that your server is up 24 hours a day. The tool will ping, traceroute and attempt to access your site via http on a regular basis to verify that your server is up. For a small fee, the service monitors your servers constantly, and contacts you by your choice of pager, cell phone or e-mail when your server goes down. The tool will also generate specialized performance statistics in real-time so that you can monitor outage patterns to ensure you're getting quality uptime from your host.

You should also routinely attempt to check server response from your own computer. If you are using a regular 56k dial-up connection, you should attempt to pull up sites located with your prospective host during peak and non-peak hours. A battery of low-cost tests is available on the network layer level of your operating system. You can test a potential hosts' network and server responsiveness from your MS-DOS or UNIX line prompt. In order to obtain a true representation of the host's services, you should select Web sites on your host's network that are typical of the services they render to their normal clients.

You should thus avoid testing the host's main Web site or premier customers. These sites are mission-critical to a hosting firm and thus are afforded an extremely high level of maintenance, which is not always representative of typical service.

In order to locate a typical client of your prospective host, execute a "whois" search. Whois is an application that looks up critical information about any Internet domain. This information includes ownership, location of the host, and most importantly, its block of network numbers. By executing the "whois -a yourhost.com" command at a UNIX line prompt, you can search your potential host's entire block of network numbers, and seek out a normal customer who is hosted on an individual network address. The customer that you use should have the approximate services that you seek. Use the ping and traceroute commands from either your UNIX or DOS prompt to test server responsiveness. You also can obtain many free or shareware WYSIWYG (what-you-see-is-what-you-get) network tools for the Windows platform that can test server responsiveness. An excellent suite of bundled network tools is provided free-of-charge by PCS Network Tools.

Using a line-prompt or WYSIWYG application, attempt to "ping" sites from the prospective host on your computer. Ping is the networking equivalent of sonar. The network tool is used to verify that a given server is actually reachable, and measures the delay that occurs when sending a data packet to it and back again.

Executing a "traceroute" from your computer is also an interesting and informative experiment to run on a hosting company. Traceroute applications allow you to map the direction that data travels over the Internet. By conducting a traceroute, you can determine whether the data you have requested from your prospective host will take a direct or indirect path to you. The most successful incident of a traceroute is therefore when data takes the shortest route to your computer.

These tests, conducted manually on a regular 56k connection will give you a rough indication of your client's response time if you were to choose the prospective host that your testing. In essence, these tests determine whether a host provides the lowest level of network latency, ensuring that data is passed to browsers and other Internet applications as quickly as possible. Your aim must be to ensure that the delay between request and response from a prospective hosting service is as short as possible. Making this determination is only possible if you conduct serious tests on prospective hosts before hosting your content there.