The "HTTP Request Count: 1-Day Average" column shows the average number of HTTP requests for one machine during a one-day period. However, not all HTTP requests are equal. When you compare the one-day average numbers with the one-hour averages in the next column, you are really getting an overall indication of the workload on the server. (Remember that the user load helps you determine the best HTTP threads setting.) To truly understand the workload, you should also rely on other system metrics (such as the CPU, response time, memory, and disk I/O). In the third HTTP Request Count column, we multiplied the average hourly rate by 24 to see how close it came to the daily request rate in the first HTTP Request Count column. This was one way that we double-checked our data.

The Server UpTime column shows results for our availability analysis. We kept track of the percentage of time that the HTTP server was available for responding to our Probe workload. For example, in the first monitoring period, the server responded to 353 of the 381 probes, or 93 percent.

Date	Server	HTTP Threads	HTTP Request Count: 1-Day Average	HTTP Request Count: 1-Hr Average	HTTP Request Count: 1-Hr Average * 24	Server UpTime
1/22	Server 2, R4.6	40	332,634	13,049	313,176	353/381 (93%)
	Server 4, R5 Beta	40	365,149	15,784	378,816	363/381 (95%)
1/26	Server 4, R5 Beta	40	205,848	16,547	397,128	473/551 (86%)
2/1	Server 4, R5 Beta	40	251,063	16,684	400,416	627/664 (94%)
2/15	Server 2, R4.6	80	518,947	25,860	620,640	590/594 (99%)
	Server 5, R5 Beta	80	450,644	22,820	547,680	414/594 (70%)
2/22	Server 2, R4.6	80	566,959	24,800	595,200	639/642 (99.5%)
3/1	Server 2, R4.6	80	413,203	16,264	390,336	210/211 (99.5%)
	Server 2, R4.6	120	240,642	13,257	318,168	414/458 (90%)
	Server 5, R5 Beta	80	382,125	17,228	413,472	273/280 (97.5%)