Initial development of the testing engine has been completed. The following tests are currently running and collecting data:

• Connection Time. The time to make a TCP/IP connection on port 80 (The Web) is measured in this test. Currently this test is running in an automated fashion every 15 minutes.
• Static Page Retrieval. The time to retrieve a static HTML web page is measured. Connection time is excluded and the page used for this test is standardized for all hosts tested. The size of the retrieved file is verified to ensure the accuracy of the test. This test is currently being run manually. The usefulness of this test remains to be determined. At the present time results do not appear to establish any relevant differences between hosts.
• Dynamic Page Retrieval. The time to retrieve a dynamic PHP database driven web page is measured. The content of the page is generated via MySQL calls. This standardized test page is based on a typical CMS page. Connection time is excluded in this test. Due to the complexity of this test, it is only being performed manually and the dataset will be regularly flushed until the testing parameters are finalized.
• Large File Retrieval. The time to retrieve a large static file is measured. Connection time is excluded and the file is approximately 10 MB in size. Out of respect for bandwidth, this test is being run manually, however results will begin to be collected in an automated fashion in the near future.

The main testing engine is currently collecting data on a small number of hosts and undergoing evaluation. It is logging data on server TCP connection time and gross bandwidth rates for typical dynamic web pages and static data files.

After a (best guess) statistically significant number of samples are taken the results will be evaluated for testing engine consistency and margin of error. The main difficulty is evaluating and eliminating factors that skew results such as temporal testing node performance. After all if a node is not performing consistently the results it returns will also not be consistent. It may become necessary to add in a benchmarking feature to ensure more accurate and directly comparable results. A less preferred option would be to average a greater number of samples

Once the general trustworthiness and margin of error are determined, the system will be put into production and results will start to be shown at this site after a sufficient amount of data is collected. If operation is shown to be reliable and accurate, data will then be processed and published to this site in real time.

The bottom line is that you will be able to directly compare speeds and reliability of each hosting provider reviewed. The best assurance of the quality of a web hosting provider is seeing the real world performance.

A review of Sunwave shared hosting has been published and updated. The review is based on more than 2 years of experience with the host.

Sunwave is not a well known provider, but might be worth consideration. The availability of a no pre-pay $7/mo plan with dedicated IP addresses may make up for the lack of 24/7 support.

Please see the full review:
Sunwave Review