Package com.github.benmanes.caffeine.cache.simulator

Class Synthetic

java.lang.Object

com.github.benmanes.caffeine.cache.simulator.Synthetic

public final class Synthetic
extends Object

A generator of synthetic cache events to simulate different caching patterns.

Method Summary

Modifier and Type	Method	Description
static LongStream	counter(int start, int events)	Returns a sequence of unique integers.
static LongStream	exponential(double mean, int events)	Returns a sequence of events based on an exponential distribution.
static TraceReader.KeyOnlyTraceReader	generate(BasicSettings.TraceSettings settings)	Returns a sequence of events based on the setting's distribution.
static LongStream	hotspot(int lowerBound, int upperBound, double hotsetFraction, double hotOpnFraction, int events)	Returns a sequence of events resembling a hotspot distribution where x% of operations access y% of data items.
static LongStream	repeating(int items, int events)	Returns a repeating sequence of integers.
static LongStream	scrambledZipfian(int items, double constant, int events)	Returns a sequence of events where some items are more popular than others, according to a zipfian distribution.
static LongStream	skewedZipfianLatest(int items, int events)	Returns a zipfian sequence with a popularity distribution of items, skewed to favor recent items significantly more than older items.
static LongStream	uniform(int lowerBound, int upperBound, int events)	Returns a sequence of events where items are selected uniformly randomly from the interval inclusively.
static LongStream	zipfian(int items, double constant, int events)	Returns a sequence of events where some items are more popular than others, according to a zipfian distribution.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Method Detail

generate

public static TraceReader.KeyOnlyTraceReader generate(BasicSettings.TraceSettings settings)

Returns a sequence of events based on the setting's distribution.

counter

public static LongStream counter(int start,
                                 int events)

Returns a sequence of unique integers.

Parameters:

start - the number that the counter starts from

events - the number of events in the distribution

repeating

public static LongStream repeating(int items,
                                   int events)

Returns a repeating sequence of integers.

Parameters:

items - the number of items in the distribution

events - the number of events in the distribution

uniform

public static LongStream uniform(int lowerBound,
                                 int upperBound,
                                 int events)

Returns a sequence of events where items are selected uniformly randomly from the interval inclusively.

Parameters:

lowerBound - lower bound of the distribution

upperBound - upper bound of the distribution

events - the number of events in the distribution

Returns:

a stream of cache events

exponential

public static LongStream exponential(double mean,
                                     int events)

Returns a sequence of events based on an exponential distribution. Smaller intervals are more frequent than larger ones, and there is no bound on the length of an interval.

Parameters:

mean - mean arrival rate of gamma (a half life of 1/gamma)

events - the number of events in the distribution

hotspot

public static LongStream hotspot(int lowerBound,
                                 int upperBound,
                                 double hotsetFraction,
                                 double hotOpnFraction,
                                 int events)

Returns a sequence of events resembling a hotspot distribution where x% of operations access y% of data items. The parameters specify the bounds for the numbers, the percentage of the interval which comprises the hot set and the percentage of operations that access the hot set. Numbers of the hot set are always smaller than any number in the cold set. Elements from the hot set and the cold set are chosen using a uniform distribution.

Parameters:

lowerBound - lower bound of the distribution

upperBound - upper bound of the distribution

hotsetFraction - percentage of data item

hotOpnFraction - percentage of operations accessing the hot set

events - the number of events in the distribution

scrambledZipfian

public static LongStream scrambledZipfian(int items,
                                          double constant,
                                          int events)

Returns a sequence of events where some items are more popular than others, according to a zipfian distribution. Unlike zipfian(int, double, int), the generated sequence scatters the "popular" items across the item space. Use if you don't want the head of the distribution (the popular items) clustered together.

Parameters:

items - the number of items in the distribution

constant - the skew factor for the distribution

events - the number of events in the distribution

skewedZipfianLatest

public static LongStream skewedZipfianLatest(int items,
                                             int events)

Returns a zipfian sequence with a popularity distribution of items, skewed to favor recent items significantly more than older items.

Parameters:

items - the number of items in the distribution

events - the number of events in the distribution

zipfian

public static LongStream zipfian(int items,
                                 double constant,
                                 int events)

Returns a sequence of events where some items are more popular than others, according to a zipfian distribution.

Parameters:

items - the number of items in the distribution

constant - the skew factor for the distribution

events - the number of events in the distribution