To generate performance data, we insert randomly generated objects into a bag and then query for them using various methods. Each object contains: (1) a "color" field randomly selected from 10 color names, (2) a "phase" field selected from the integers 1-4, (3) a "score" field randomly generated by a double precision number between 0.0 and 1.0, and (4) a "description" field composed of 200-450 characters of random text.

For each combination of query parameters, we run 20 iterations. On each iteration, we randomly select the constraint values and a limit of either 5, 10, or 20."range" constraints are selected with a min and a max randomly selected between 0.0 and 1.0. "in" constraints are selected with 2-5 random values.

number of objects in bag	'color' constraint	'score' constraint	mean duration(ms)
10000	none	none	6
10000	none	range	16
10000	none	sort	10
10000	is	none	5
10000	is	range	12
10000	is	sort	60
10000	in	none	5
10000	in	range	10
10000	in	sort	50
10000	sort	none	7
10000	sort	range	17
100000	none	none	8
100000	none	range	10
100000	none	sort	7
100000	is	none	6
100000	is	range	10
100000	is	sort	67
100000	in	none	4
100000	in	range	10
100000	in	sort	38
100000	sort	none	6
100000	sort	range	13
1000000	none	none	7
1000000	none	range	1217
1000000	none	sort	35
1000000	is	none	11
1000000	is	range	541
1000000	is	sort	485
1000000	in	none	6
1000000	in	range	327
1000000	in	sort	144
1000000	sort	none	6
1000000	sort	range	147