Package brave.sampler

Class Sampler

java.lang.Object

brave.sampler.Sampler

Direct Known Subclasses:

BoundarySampler, CountingSampler, RateLimitingSampler

public abstract class Sampler
extends Object

Sampler is responsible for deciding if a particular trace should be "sampled", i.e. whether the overhead of tracing will occur and/or if a trace will be reported to the collection tier.

Zipkin v1 uses before-the-fact sampling. This means that the decision to keep or drop the trace is made before any work is measured, or annotations are added. As such, the input parameter to zipkin v1 samplers is the trace ID (lower 64-bits under the assumption all bits are random).

The instrumentation sampling decision happens once, at the root of the trace, and is propagated downstream. For this reason, the algorithm needn't be consistent based on trace ID.

Field Summary

Fields

Modifier and Type	Field	Description
static Sampler	ALWAYS_SAMPLE
static Sampler	NEVER_SAMPLE

Constructor Summary

Constructors

Constructor	Description
Sampler()

Method Summary

Modifier and Type	Method	Description
static Sampler	create(float probability)	Returns a sampler, given a probability expressed as a percentage.
abstract boolean	isSampled(long traceId)	Returns true if the trace ID should be measured.

Methods inherited from class java.lang.Object

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

Field Details

ALWAYS_SAMPLE

public static final Sampler ALWAYS_SAMPLE

NEVER_SAMPLE

public static final Sampler NEVER_SAMPLE

Constructor Details

Sampler

public Sampler()

Method Details

isSampled

public abstract boolean isSampled(long traceId)

Returns true if the trace ID should be measured.

Parameters:

traceId - The trace ID to be decided on, can be ignored

create

public static Sampler create(float probability)

Returns a sampler, given a probability expressed as a percentage.

The sampler returned is good for low volumes of traffic (<100K requests), as it is precise. If you have high volumes of traffic, consider BoundarySampler.

Parameters:

probability - probability a trace will be sampled. minimum is 0.01, or 1% of traces