[wiki.stajistics] push by - Edited wiki page GettingStarted through web user interface. on 2011-06-21 03:44 GMT
staji...@googlecode.com
Author: troy.k...@gmail.com
Date: Mon Jun 20 20:44:31 2011
Log: Edited wiki page GettingStarted through web user interface.
http://code.google.com/p/stajistics/source/detail?repo=wiki&r=493cd0522afc
Modified:
/GettingStarted.wiki
=======================================
--- /GettingStarted.wiki Mon Jun 20 20:20:47 2011
+++ /GettingStarted.wiki Mon Jun 20 20:44:31 2011
@@ -52,6 +52,7 @@
The following is a table of tracker implementations that are standard in
the stajistics-core module.
|| Tracker Name || Description ||
+|| IncidentTracker || Records single occurrences of events. ||
|| TimeDurationTracker || Records the amount of time, in various
precisions, between a start and end point. ||
|| ConcurrentAccessTracker || Tracks how many simultaneous accesses occur
to a span defined by a start and end point. ||
|| HitFrequencyTracker || Measures the amount of time between accesses to
a span defined by a start and end point. ||
@@ -59,9 +60,20 @@
|| ThreadCPUTimeTracker || Records the amount of CPU time allotted to a
thread between a start and end point. ||
|| ThreadBlockTimeTracker || Measures the amount of time a thread spent
blocking between a start and end point. ||
|| ThreadWaitTimeTracker || Measures the amount of time a thread spent
waiting between a start and end point. ||
+|| ManualTracker || Collects whatever numerical data is fed to it. ||
|| CompositeSpanTracker || Aggregates several configurable trackers into
one for simple collection of multiple kinds of data on a single span
defined by a start and end point. ||
-TODO: talk about where the data goes.
+== What Gets Collected? ==
+
+When a tracker collects a unit of statistical data, it hands it off to the
session where it is stored and/or minor initial calculations are performed
upon it. A piece of data stored in a session is known as a field, and is
identified by a name.
+
+By default, a session will only keep a handful of calculation results that
can be derived from streams of data points; it will not store the data
points themselves. For example, given a data point time-spent-in-block-A,
only the sum and average (among other things) are stored, not the
individual times themselves. Why? Storing all data points in a stream would
have high demands on memory and other resources. If a full history of data
points is desired, Stajistics can support that through custom DataRecorders.
+
+The following is a table of standard fields stored in a session:
+
+|| Field Name || Example Value || Description ||
+TODO
+
== Track Something! ==