These statistics are always output.
========================================================================= == Summary information ========================================================================= First date analyzed: Thu Nov 3 14:12:20 2005 Last date analyzed: Thu Dec 15 16:04:52 2005 Time period analyzed: 42d 1h 52m 32s Lines analyzed, total: 383545 Valid requests in time period: 383391 Avg ops/sec: 0.105 Bytes/sec: 2238.883 Mb/sec: 0.017 Max ops in one second: 22 Max ops in five seconds: 75 (15.000 ops/sec) Interpolated one-second ops avg: 1.796 ops/sec Interpolated five-second ops avg: 1.110 ops/sec ----- Round-trip Response Time, ms ----- Requests Analyzed: 383391 Average: 345.183 Standard Deviation: 561.897 Maximum: 40077.56 Minimum: 1.55 Over Threshold (1250): 11269 (2.939%) ----- MDEX Engine-only Processing Time, ms ----- Requests Analyzed: 383391 Average: 48.861 Standard Deviation: 132.229 Maximum: 15475.75 Minimum: 0.00 Over Threshold (750): 1325 (0.346%) ----- Response Differential, ms (round-trip minus engine-only) ----- Requests Analyzed: 383391 Average: 296.321 Standard Deviation: 544.660 Maximum: 40003.33 Minimum: 1.3 Over Threshold (500): 42730 (11.145%) ----- Response Size, bytes ----- Requests Analyzed: 383391 Average: 21230.479 Standard Deviation: 49118.382 Maximum: 895065 Minimum: 0 Over Threshold (393216): 456 (0.119%)
This section outputs information about the performance and characteristics of the MDEX Engine as a whole.
The following statistics are output in this section:
First date analyzed: the timestamp of the earliest request analyzed within the logs. Translation of the time recorded in the log to human-readable time is controlled with the
--hourOffsetflag.Last date analyzed: the timestamp of the latest request analyzed within the logs. Translation of the time recorded in the log to human-readable time is controlled with the
--hourOffsetflag.Time period analyzed: the timespan between the earliest analyzed request and the latest analyzed request.
Lines analyzed, total: the number of requests inspected by the Request Log Analyzer within this set of logfiles. Not all of these requests will have been analyzed. MDEX Engine startup and shutdown status messages write a line to the logfile and thus contribute to this metric but are otherwise ignored by the Request Log Analyzer. Other lines are controlled by the --ignore* flags.
Valid requests in time period: the number of requests actually analyzed by the Request Log Analyzer within this set of logfiles.
Avg ops/sec: the achieved throughput, in operations (queries) per second. This metric is calculated by dividing the number of requests analyzed by the number of seconds in the time period analyzed. Note that this metric can be heavily skewed by large stretches of time that contain no requests.
Bytes/sec: the achieved network bandwidth usage, in bytes. This metric is calculated by dividing the sum of all data packet sizes analyzed by the number of seconds in the time period analyzed. Note that this metric can be heavily skewed by large stretches of time that contain no requests.
Max ops in one second: the highest achieved throughput metric for a one-second timespan. The Request Log Analyzer slices each logfile into one-second pieces, finds the one-second timespan that contains the most analyzed requests, and reports that number here.
Max ops in five seconds: the highest achieved throughput metric for a five-second timespan. The Request Log Analyzer slices each logfile into five-second pieces, finds the five-second timespan that contains the most analyzed requests, and reports the throughput of that timespan here. Note that the slicing algorithm is arbitrary; it is possible that the last two seconds of a slice plus the next three seconds of the following consecutive slice contain more requests than any of the arbitrary slices.
Interpolated one-second ops avg: a version of achieved throughput that attempts to address the weaknesses of the previous achieved throughput measurement. This metric is calculated by dividing the number of requests analyzed by the number of seconds in the logs that actually contain requests. For most logfiles this metric will be a more accurate achieved throughput. For logfiles containing a large number of requests that required more than one second to process, this metric will be inaccurate.
Interpolated five-second ops avg: a version of achieved throughput that attempts to address the weaknesses of the previous achieved throughput measurement. This metric is calculated by dividing the number of requests analyzed by the number of five-second time slices in the logs that actually contain requests, then adjusting by 5. For most logfiles this metric will be a more accurate achieved throughput. For logfiles containing a large number of requests that required more than one second to process, this metric will be inaccurate.
Round-trip Response Time, ms: The round-trip response time for all analyzed requests. See Common Metrics for an explanation of round-trip response time.
MDEX Engine-only Processing Time, ms: The engine-only processing time for all analyzed requests. See Common Metrics for an explanation of engine-only processing time.
Response Differential, ms (round-trip minus engine-only): The response differential for all analyzed requests. The response differential is measured as the difference between the round-trip response time and the engine-only processing time. This metric measures how long requests spend in the request queue and the response queue combined.
Response Size, bytes: The response size for all analyzed requests. See Common Metrics for an explanation of round-trip response size.
