beam

package
v2.24.0-RC1+incompatible Latest Latest
Warning

This package is not in the latest version of its module.

 Go to latest Published: Aug 22, 2020 License: Apache-2.0 Imports: 20 Imported by: 0

Documentation ¶

Overview ¶

Package beam is an implementation of the Apache Beam (https://beam.apache.org) programming model in Go. Beam provides a simple, powerful model for building both batch and streaming parallel data processing pipelines.

For more on the Beam model see: https://beam.apache.org/documentation/programming-guide

For design choices this implementation makes see: https://s.apache.org/beam-go-sdk-design-rfc

Example (GettingStarted) ¶ 
Output:
Example (MetricsDeclaredAnywhere) ¶ 
Output:

PTransformID: "example"
	example.namespace.characters - count: 13 sum: 43 min: 2 max: 5
	example.namespace.count - value: 13
Example (MetricsReusable) ¶ 
Output:

PTransformID: "extract1"
	example.reusable.count - value: 6
PTransformID: "extract2"
	example.reusable.count - value: 14

Index ¶

Examples ¶

Constants ¶

This section is empty.

Variables ¶

View Source 
var (
	TType = typex.TType
	UType = typex.UType
	VType = typex.VType
	WType = typex.WType
	XType = typex.XType
	YType = typex.YType
	ZType = typex.ZType
)

These are the reflect.Type instances of the universal types, which are used when binding actual types to "generic" DoFns that use Universal Types.

View Source 
var EventTimeType = typex.EventTimeType

EventTimeType is the reflect.Type of EventTime.

View Source 
var PipelineOptions = runtime.GlobalOptions

PipelineOptions are global options for the active pipeline. Options can be defined any time before execution and are re-created by the harness on remote execution workers. Global options should be used sparingly.

Functions ¶

func Init ¶ 

func Init()

Init is the hook that all user code must call after flags processing and other static initialization, for now.

func Initialized ¶ 

func Initialized() bool

Initialized exposes the initialization status for runners.

func NewPipelineWithRoot ¶ 

func NewPipelineWithRoot() (*Pipeline, Scope)

NewPipelineWithRoot creates a new empty pipeline and its root scope.

func ParDo0 ¶ 

func ParDo0(s Scope, dofn interface{}, col PCollection, opts ...Option)

ParDo0 inserts a ParDo with zero output transform into the pipeline.

func ParDo2 ¶ 

func ParDo2(s Scope, dofn interface{}, col PCollection, opts ...Option) (PCollection, PCollection)

ParDo2 inserts a ParDo with 2 outputs into the pipeline.

func ParDo3 ¶ 

func ParDo3(s Scope, dofn interface{}, col PCollection, opts ...Option) (PCollection, PCollection, PCollection)

ParDo3 inserts a ParDo with 3 outputs into the pipeline.

func ParDo4 ¶ 

func ParDo4(s Scope, dofn interface{}, col PCollection, opts ...Option) (PCollection, PCollection, PCollection, PCollection)

ParDo4 inserts a ParDo with 4 outputs into the pipeline.

func ParDo5 ¶ 

func ParDo5(s Scope, dofn interface{}, col PCollection, opts ...Option) (PCollection, PCollection, PCollection, PCollection, PCollection)

ParDo5 inserts a ParDo with 5 outputs into the pipeline.

func ParDo6 ¶ 

func ParDo6(s Scope, dofn interface{}, col PCollection, opts ...Option) (PCollection, PCollection, PCollection, PCollection, PCollection, PCollection)

ParDo6 inserts a ParDo with 6 outputs into the pipeline.

func ParDo7 ¶ 

func ParDo7(s Scope, dofn interface{}, col PCollection, opts ...Option) (PCollection, PCollection, PCollection, PCollection, PCollection, PCollection, PCollection)

ParDo7 inserts a ParDo with 7 outputs into the pipeline.

func RegisterCoder ¶ 

func RegisterCoder(t reflect.Type, encoder, decoder interface{})

RegisterCoder registers a user defined coder for a given type, and will be used if there is no existing beam coder for that type. Must be called prior to beam.Init(), preferably in an init() function.

The coder used for a given type follows this ordering:

  1. Coders for Known Beam types.
  2. Coders registered for specific types
  3. Coders registered for interfaces types
  4. Default coder (JSON)

Coders for interface types are iterated over to check if a type satisfies them, and the most recent one registered will be used.

Repeated registrations of the same type overrides prior ones.

RegisterCoder additionally registers the type, and coder functions as per RegisterType and RegisterFunction to avoid redundant calls.

Supported Encoder Signatures 

func(T) []byte
func(reflect.Type, T) []byte
func(T) ([]byte, error)
func(reflect.Type, T) ([]byte, error)

Supported Decoder Signatures 

func([]byte) T
func(reflect.Type, []byte) T
func([]byte) (T, error)
func(reflect.Type, []byte) (T, error)

where T is the matching user type.

func RegisterFunction ¶ 

func RegisterFunction(fn interface{})

RegisterFunction allows function registration. It is beneficial for performance and is needed for functions -- such as custom coders -- serialized during unit tests, where the underlying symbol table is not available. It should be called in init() only. Returns the external key for the function.

func RegisterInit ¶ 

func RegisterInit(hook func())

RegisterInit registers an Init hook. Hooks are expected to be able to figure out whether they apply on their own, notably if invoked in a remote execution environment. They are all executed regardless of the runner.

func RegisterRunner ¶ 

func RegisterRunner(name string, fn func(ctx context.Context, p *Pipeline) error)

RegisterRunner associates the name with the supplied runner, making it available to execute a pipeline via Run.

func RegisterType ¶ 

func RegisterType(t reflect.Type)

RegisterType inserts "external" types into a global type registry to bypass serialization and preserve full method information. It should be called in init() only. TODO(wcn): the canonical definition of "external" is in v1.proto. We need user facing copy for this important concept.

func Run ¶ 

func Run(ctx context.Context, runner string, p *Pipeline) error

Run executes the pipeline using the selected registred runner. It is customary to define a "runner" with no default as a flag to let users control runner selection.

func ValidateKVType ¶ 

func ValidateKVType(col PCollection) (typex.FullType, typex.FullType)

ValidateKVType panics if the type of the PCollection is not KV<A,B>. It returns (A,B).

func ValidateNonCompositeType ¶ 

func ValidateNonCompositeType(col PCollection) typex.FullType

ValidateNonCompositeType panics if the type of the PCollection is not a composite type. It returns the type.

Types ¶

type Coder ¶ 

type Coder struct {
	// contains filtered or unexported fields
}

Coder defines how to encode and decode values of type 'A' into byte streams. Coders are attached to PCollections of the same type. For PCollections consumed by GBK, the attached coders are required to be deterministic.

func DecodeCoder ¶ 

func DecodeCoder(data string) (Coder, error)

DecodeCoder decodes a coder. Any custom coder function symbol must be resolvable via the runtime.GlobalSymbolResolver. The types must be encodable.

func NewCoder ¶ 

func NewCoder(t FullType) Coder

NewCoder infers a Coder for any bound full type.

func (Coder) IsValid ¶ 

func (c Coder) IsValid() bool

IsValid returns true iff the Coder is valid. Any use of an invalid Coder will result in a panic.

func (Coder) String ¶ 

func (c Coder) String() string

func (Coder) Type ¶ 

func (c Coder) Type() FullType

Type returns the full type 'A' of elements the coder can encode and decode. 'A' must be a concrete full type, such as int or KV<int,string>.

type Counter ¶ 

type Counter struct {
	*metrics.Counter
}

Counter is a metric that can be incremented and decremented, and is aggregated by the sum.

Counters are safe to use in multiple bundles simultaneously, but not generally threadsafe. Your DoFn needs to manage the thread safety of Beam metrics for any additional concurrency it uses.

func NewCounter ¶ 

func NewCounter(namespace, name string) Counter

NewCounter returns the Counter with the given namespace and name.

func (Counter) Dec ¶ 

func (c Counter) Dec(ctx context.Context, v int64)

Dec decrements the counter within by the given amount.

Example ¶ 
Output:

func (Counter) Inc ¶ 

func (c Counter) Inc(ctx context.Context, v int64)

Inc increments the counter within by the given amount.

Example ¶ 
Output:

type Distribution ¶ 

type Distribution struct {
	*metrics.Distribution
}

Distribution is a metric that records various statistics about the distribution of reported values.

Distributions are safe to use in multiple bundles simultaneously, but not generally threadsafe. Your DoFn needs to manage the thread safety of Beam metrics for any additional concurrency it uses.

func NewDistribution ¶ 

func NewDistribution(namespace, name string) Distribution

NewDistribution returns the Distribution with the given namespace and name.

func (Distribution) Update ¶ 

func (c Distribution) Update(ctx context.Context, v int64)

Update adds an observation to this distribution.

Example ¶ 
Output:

type ElementDecoder ¶ 

type ElementDecoder = coder.ElementDecoder

ElementDecoder encapsulates being able to decode an element from a reader.

func NewElementDecoder ¶ 

func NewElementDecoder(t reflect.Type) ElementDecoder

NewElementDecoder returns an ElementDecoder the given type.

type ElementEncoder ¶ 

type ElementEncoder = coder.ElementEncoder

ElementEncoder encapsulates being able to encode an element into a writer.

func NewElementEncoder ¶ 

func NewElementEncoder(t reflect.Type) ElementEncoder

NewElementEncoder returns a new encoding function for the given type.

type EncodedCoder ¶ 

type EncodedCoder struct {
	// Coder is the coder to preserve across serialization.
	Coder Coder
}

EncodedCoder is a serialization wrapper around a coder for convenience.

func (EncodedCoder) MarshalJSON ¶ 

func (w EncodedCoder) MarshalJSON() ([]byte, error)

MarshalJSON returns the JSON encoding this value.

func (*EncodedCoder) UnmarshalJSON ¶ 

func (w *EncodedCoder) UnmarshalJSON(buf []byte) error

UnmarshalJSON sets the state of this instance from the passed in JSON.

type EncodedFunc ¶ 

type EncodedFunc struct {
	// Fn is the function to preserve across serialization.
	Fn reflectx.Func
}

EncodedFunc is a serialization wrapper around a function for convenience.

func (EncodedFunc) MarshalJSON ¶ 

func (w EncodedFunc) MarshalJSON() ([]byte, error)

MarshalJSON returns the JSON encoding this value.

func (*EncodedFunc) UnmarshalJSON ¶ 

func (w *EncodedFunc) UnmarshalJSON(buf []byte) error

UnmarshalJSON sets the state of this instance from the passed in JSON.

type EncodedType ¶ 

type EncodedType struct {
	// T is the type to preserve across serialization.
	T reflect.Type
}

EncodedType is a serialization wrapper around a type for convenience.

func (EncodedType) MarshalJSON ¶ 

func (w EncodedType) MarshalJSON() ([]byte, error)

MarshalJSON returns the JSON encoding this value.

func (*EncodedType) UnmarshalJSON ¶ 

func (w *EncodedType) UnmarshalJSON(buf []byte) error

UnmarshalJSON sets the state of this instance from the passed in JSON.

type EventTime ¶ 

type EventTime = typex.EventTime

EventTime represents the time of the event that generated an element. This is distinct from the time when an element is processed.

type FullType ¶ 

type FullType = typex.FullType

FullType represents the tree structure of data types processed by the graph. It allows representation of composite types, such as KV<int, string> or CoGBK<int, int>, as well as "generic" such types, KV<int,T> or CoGBK<X,Y>, where the free "type variables" are the fixed universal types: T, X, etc.

type Gauge ¶ 

type Gauge struct {
	*metrics.Gauge
}

Gauge is a metric that can have its new value set, and is aggregated by taking the last reported value.

Gauge are safe to use in multiple bundles simultaneously, but not generally threadsafe. Your DoFn needs to manage the thread safety of Beam metrics for any additional concurrency it uses.

func NewGauge ¶ 

func NewGauge(namespace, name string) Gauge

NewGauge returns the Gauge with the given namespace and name.

func (Gauge) Set ¶ 

func (c Gauge) Set(ctx context.Context, v int64)

Set sets the current value for this gauge.

Example ¶ 
Output:

type Option ¶ 

type Option interface {
	// contains filtered or unexported methods
}

Option is an optional value or context to a transformation, used at pipeline construction time. The primary use case is providing side inputs.

type PCollection ¶ 

type PCollection struct {
	// contains filtered or unexported fields
}

PCollection is an immutable collection of values of type 'A', which must be a concrete type, such as int or KV<int,string>. A PCollection can contain either a bounded or unbounded number of elements. Bounded and unbounded PCollections are produced as the output of PTransforms (including root PTransforms like textio.Read), and can be passed as the inputs of other PTransforms. Some root transforms produce bounded PCollections and others produce unbounded ones.

Each element in a PCollection has an associated timestamp. Sources assign timestamps to elements when they create PCollections, and other PTransforms propagate these timestamps from their input to their output implicitly or explicitly.

Additionally, each element is assigned to a set of windows. By default, all elements are assigned into a single default window, GlobalWindow.

func AddFixedKey ¶ 

func AddFixedKey(s Scope, col PCollection) PCollection

AddFixedKey adds a fixed key (0) to every element.

func CoGroupByKey ¶ 

func CoGroupByKey(s Scope, cols ...PCollection) PCollection

CoGroupByKey inserts a CoGBK transform into the pipeline.

func Combine ¶ 

func Combine(s Scope, combinefn interface{}, col PCollection, opts ...Option) PCollection

Combine inserts a global Combine transform into the pipeline. It expects a PCollection<T> as input where T is a concrete type. Combine supports TypeDefinition options for binding generic types in combinefn.

func CombinePerKey ¶ 

func CombinePerKey(s Scope, combinefn interface{}, col PCollection, opts ...Option) PCollection

CombinePerKey inserts a GBK and per-key Combine transform into the pipeline. It expects a PCollection<KV<K,T>>. The CombineFn may optionally take a key parameter. CombinePerKey supports TypeDefinition options for binding generic types in combinefn.

func Create ¶ 

func Create(s Scope, values ...interface{}) PCollection

Create inserts a fixed non-empty set of values into the pipeline. The values must be of the same type 'A' and the returned PCollection is of type A.

The returned PCollections can be used as any other PCollections. The values are JSON-coded. Each runner may place limits on the sizes of the values and Create should generally only be used for small collections.

Example ¶ 
Output:

func CreateList ¶ 

func CreateList(s Scope, list interface{}) PCollection

CreateList inserts a fixed set of values into the pipeline from a slice or array. Unlike Create this supports the creation of an empty PCollection.

Example ¶ 
Output:

func DropKey ¶ 

func DropKey(s Scope, col PCollection) PCollection

DropKey drops the key for an input PCollection<KV<A,B>>. It returns a PCollection<B>.

func DropValue ¶ 

func DropValue(s Scope, col PCollection) PCollection

DropValue drops the value for an input PCollection<KV<A,B>>. It returns a PCollection<A>.

func Explode ¶ 

func Explode(s Scope, col PCollection) PCollection

Explode is a PTransform that takes a single PCollection<[]A> and returns a PCollection<A> containing all the elements for each incoming slice.

Example ¶ 
Output:

func External ¶ 

func External(s Scope, spec string, payload []byte, in []PCollection, out []FullType, bounded bool) []PCollection

External defines a Beam external transform. The interpretation of this primitive is runner specific. The runner is responsible for parsing the payload based on the spec provided to implement the behavior of the operation. Transform libraries should expose an API that captures the user's intent and serialize the payload as a byte slice that the runner will deserialize.

func Flatten ¶ 

func Flatten(s Scope, cols ...PCollection) PCollection

Flatten is a PTransform that takes either multiple PCollections of type 'A' and returns a single PCollection of type 'A' containing all the elements in all the input PCollections. The name "Flatten" suggests taking a list of lists and flattening them into a single list.

By default, the Coder of the output PCollection is the same as the Coder of the first PCollection.

Example ¶ 
Output:

func GroupByKey ¶ 

func GroupByKey(s Scope, a PCollection) PCollection

GroupByKey is a PTransform that takes a PCollection of type KV<A,B>, groups the values by key and windows, and returns a PCollection of type GBK<A,B> representing a map from each distinct key and window of the input PCollection to an iterable over all the values associated with that key in the input per window. Each key in the output PCollection is unique within each window.

GroupByKey is analogous to converting a multi-map into a uni-map, and related to GROUP BY in SQL. It corresponds to the "shuffle" step between the Mapper and the Reducer in the MapReduce framework.

Two keys of type A are compared for equality by first encoding each of the keys using the Coder of the keys of the input PCollection, and then comparing the encoded bytes. This admits efficient parallel evaluation. Note that this requires that the Coder of the keys be deterministic.

By default, input and output PCollections share a key Coder and iterable values in the input and output PCollection share an element Coder.

GroupByKey is a key primitive in data-parallel processing, since it is the main way to efficiently bring associated data together into one location. It is also a key determiner of the performance of a data-parallel pipeline.

See CoGroupByKey for a way to group multiple input PCollections by a common key at once.

Example ¶ 
Output:

func Impulse ¶ 

func Impulse(s Scope) PCollection

Impulse emits a single empty []byte into the global window. The resulting PCollection is a singleton of type []byte.

The purpose of Impulse is to trigger another transform, such as ones that take all information as side inputs.

Example ¶ 
Output:

func ImpulseValue ¶ 

func ImpulseValue(s Scope, value []byte) PCollection

ImpulseValue emits the supplied byte slice into the global window. The resulting PCollection is a singleton of type []byte.

Example ¶ 
Output:

func Must ¶ 

func Must(a PCollection, err error) PCollection

Must returns the input, but panics if err != nil.

func MustN ¶ 

func MustN(list []PCollection, err error) []PCollection

MustN returns the input, but panics if err != nil.

func ParDo ¶ 

func ParDo(s Scope, dofn interface{}, col PCollection, opts ...Option) PCollection

ParDo is the core element-wise PTransform in Apache Beam, invoking a user-specified function on each of the elements of the input PCollection to produce zero or more output elements, all of which are collected into the output PCollection. Use one of the ParDo variants for a different number of output PCollections. The PCollections do no need to have the same types.

Elements are processed independently, and possibly in parallel across distributed cloud resources. The ParDo processing style is similar to what happens inside the "Mapper" or "Reducer" class of a MapReduce-style algorithm.

DoFns ¶

The function to use to process each element is specified by a DoFn, either as single function or as a struct with methods, notably ProcessElement. The struct may also define Setup, StartBundle, FinishBundle and Teardown methods. The struct is JSON-serialized and may contain construction-time values.

Conceptually, when a ParDo transform is executed, the elements of the input PCollection are first divided up into some number of "bundles". These are farmed off to distributed worker machines (or run locally, if using the direct runner). For each bundle of input elements processing proceeds as follows:

  • If a struct, a fresh instance of the argument DoFn is created on a worker from json serialization, and the Setup method is called on this instance, if present. A runner may reuse DoFn instances for multiple bundles. A DoFn that has terminated abnormally (by returning an error) will never be reused.
  • The DoFn's StartBundle method, if provided, is called to initialize it.
  • The DoFn's ProcessElement method is called on each of the input elements in the bundle.
  • The DoFn's FinishBundle method, if provided, is called to complete its work. After FinishBundle is called, the framework will not again invoke ProcessElement or FinishBundle until a new call to StartBundle has occurred.
  • If any of Setup, StartBundle, ProcessElement or FinishBundle methods return an error, the Teardown method, if provided, will be called on the DoFn instance.
  • If a runner will no longer use a DoFn, the Teardown method, if provided, will be called on the discarded instance.

Each of the calls to any of the DoFn's processing methods can produce zero or more output elements. All of the of output elements from all of the DoFn instances are included in an output PCollection.

For example: 

words := beam.ParDo(s, &Foo{...}, ...)
lengths := beam.ParDo(s, func (word string) int) {
      return len(word)
}, words)

Each output element has the same timestamp and is in the same windows as its corresponding input element. The timestamp can be accessed and/or emitted by including a EventTime-typed parameter. The name of the function or struct is used as the DoFn name. Function literals do not have stable names and should thus not be used in production code.

Side Inputs ¶

While a ParDo processes elements from a single "main input" PCollection, it can take additional "side input" PCollections. These SideInput along with the DoFn parameter form express styles of accessing PCollection computed by earlier pipeline operations, passed in to the ParDo transform using SideInput options, and their contents accessible to each of the DoFn operations. For example: 

words := ...
cufoff := ...  // Singleton PCollection<int>
smallWords := beam.ParDo(s, func (word string, cutoff int, emit func(string)) {
      if len(word) < cutoff {
           emit(word)
      }
}, words, beam.SideInput{Input: cutoff})

Additional Outputs ¶

Optionally, a ParDo transform can produce zero or multiple output PCollections. Note the use of ParDo2 to specfic 2 outputs. For example: 

words := ...
cufoff := ...  // Singleton PCollection<int>
small, big := beam.ParDo2(s, func (word string, cutoff int, small, big func(string)) {
      if len(word) < cutoff {
           small(word)
      } else {
           big(word)
      }
}, words, beam.SideInput{Input: cutoff})

By default, the Coders for the elements of each output PCollections is inferred from the concrete type.

No Global Shared State ¶

There are three main ways to initialize the state of a DoFn instance processing a bundle:

  • Define public instance variable state. This state will be automatically JSON serialized and then deserialized in the DoFn instances created for bundles. This method is good for state known when the original DoFn is created in the main program, if it's not overly large. This is not suitable for any state which must only be used for a single bundle, as DoFn's may be used to process multiple bundles.

  • Compute the state as a singleton PCollection and pass it in as a side input to the DoFn. This is good if the state needs to be computed by the pipeline, or if the state is very large and so is best read from file(s) rather than sent as part of the DoFn's serialized state.

  • Initialize the state in each DoFn instance, in a StartBundle method. This is good if the initialization doesn't depend on any information known only by the main program or computed by earlier pipeline operations, but is the same for all instances of this DoFn for all program executions, say setting up empty caches or initializing constant data.

ParDo operations are intended to be able to run in parallel across multiple worker machines. This precludes easy sharing and updating mutable state across those machines. There is no support in the Beam model for communicating and synchronizing updates to shared state across worker machines, so programs should not access any mutable global variable state in their DoFn, without understanding that the Go processes for the main program and workers will each have its own independent copy of such state, and there won't be any automatic copying of that state across Java processes. All information should be communicated to DoFn instances via main and side inputs and serialized state, and all output should be communicated from a DoFn instance via output PCollections, in the absence of external communication mechanisms written by user code.

Splittable DoFns (Experimental)

Warning: Splittable DoFns are still experimental, largely untested, and likely to have bugs.

Splittable DoFns are DoFns that are able to split work within an element, as opposed to only at element boundaries like normal DoFns. This is useful for DoFns that emit many outputs per input element and can distribute that work among multiple workers. The most common examples of this are sources.

In order to split work within an element, splittable DoFns use the concept of restrictions, which are objects that are associated with an element and describe a portion of work on that element. For example, a restriction associated with a filename might describe what byte range within that file to process. In addition to restrictions, splittable DoFns also rely on restriction trackers to track progress and perform splits on a restriction currently being processed. See the `RTracker` interface in core/sdf/sdf.go for more details.

Splitting ¶

Splitting means taking one restriction and splitting into two or more that cover the entire input space of the original one. In other words, processing all the split restrictions should produce identical output to processing the original one.

Splitting occurs in two stages. The initial splitting occurs before any restrictions have started processing. This step is used to split large restrictions into smaller ones that can then be distributed among multiple workers for processing. Initial splitting is user-defined and optional.

Dynamic splitting occurs during the processing of a restriction in runners that have implemented it. If there are available workers, runners may split the unprocessed portion of work from a busy worker and shard it to available workers in order to better distribute work. With unsplittable DoFns this can only occur on element boundaries, but for splittable DoFns this split can land within a restriction and will require splitting that restriction.

  • Note: The Go SDK currently does not support dynamic splitting for SDFs, only initial splitting. Only initially split restrictions can be distributed by liquid sharding. Stragglers will not be split during execution with dynamic splitting.

Splittable DoFn Methods ¶

Making a splittable DoFn requires the following methods to be implemented on a DoFn in addition to the usual DoFn requirements. In the following method signatures `elem` represents the main input elements to the DoFn, and should match the types used in ProcessElement. `restriction` represents the user-defined restriction, and can be any type as long as it is consistent throughout all the splittable DoFn methods:

  • `CreateInitialRestriction(element) restriction` CreateInitialRestriction creates an initial restriction encompassing an entire element. The restriction created stays associated with the element it describes.
  • `SplitRestriction(elem, restriction) []restriction` SplitRestriction takes an element and its initial restriction, and optionally performs an initial split on it, returning a slice of all the split restrictions. If no splits are desired, the method returns a slice containing only the original restriction. This method will always be called on each newly created restriction before they are processed.
  • `RestrictionSize(elem, restriction) float64` RestrictionSize returns a cheap size estimation for a restriction. This size is an abstract scalar value that represents how much work a restriction takes compared to other restrictions in the same DoFn. For example, a size of 200 represents twice as much work as a size of 100, but the numbers do not represent anything on their own. Size is used by runners to estimate work for liquid sharding.
  • `CreateTracker(restriction) restrictionTracker` CreateTracker creates and returns a restriction tracker (a concrete type implementing the `sdf.RTracker` interface) given a restriction. The restriction tracker is used to track progress processing a restriction, and to allow for dynamic splits. This method is called on each restriction right before processing begins.
  • `ProcessElement(sdf.RTracker, element, func emit(output))` For splittable DoFns, ProcessElement requires a restriction tracker before inputs, and generally requires emits to be used for outputs, since restrictions will generally produce multiple outputs. For more details on processing restrictions in a splittable DoFn, see `sdf.RTracker`.

Fault Tolerance ¶

In a distributed system, things can fail: machines can crash, machines can be unable to communicate across the network, etc. While individual failures are rare, the larger the job, the greater the chance that something, somewhere, will fail. Beam runners may strive to mask such failures by retrying failed DoFn bundles. This means that a DoFn instance might process a bundle partially, then crash for some reason, then be rerun (often as a new process) on that same bundle and on the same elements as before. Sometimes two or more DoFn instances will be running on the same bundle simultaneously, with the system taking the results of the first instance to complete successfully. Consequently, the code in a DoFn needs to be written such that these duplicate (sequential or concurrent) executions do not cause problems. If the outputs of a DoFn are a pure function of its inputs, then this requirement is satisfied. However, if a DoFn's execution has external side-effects, such as performing updates to external HTTP services, then the DoFn's code needs to take care to ensure that those updates are idempotent and that concurrent updates are acceptable. This property can be difficult to achieve, so it is advisable to strive to keep DoFns as pure functions as much as possible.

Optimization ¶

Beam runners may choose to apply optimizations to a pipeline before it is executed. A key optimization, fusion, relates to ParDo operations. If one ParDo operation produces a PCollection that is then consumed as the main input of another ParDo operation, the two ParDo operations will be fused together into a single ParDo operation and run in a single pass; this is "producer-consumer fusion". Similarly, if two or more ParDo operations have the same PCollection main input, they will be fused into a single ParDo that makes just one pass over the input PCollection; this is "sibling fusion".

If after fusion there are no more unfused references to a PCollection (e.g., one between a producer ParDo and a consumer ParDo), the PCollection itself is "fused away" and won't ever be written to disk, saving all the I/O and space expense of constructing it.

When Beam runners apply fusion optimization, it is essentially "free" to write ParDo operations in a very modular, composable style, each ParDo operation doing one clear task, and stringing together sequences of ParDo operations to get the desired overall effect. Such programs can be easier to understand, easier to unit-test, easier to extend and evolve, and easier to reuse in new programs. The predefined library of PTransforms that come with Beam makes heavy use of this modular, composable style, trusting to the runner to "flatten out" all the compositions into highly optimized stages.

See https://beam.apache.org/documentation/programming-guide/#pardo for the web documentation for ParDo

Example (AdditionalOutputs) ¶

Optionally, a ParDo transform can produce zero or multiple output PCollections. Note the use of ParDo2 to specify 2 outputs. 

Output:

func ParDoN ¶ 

func ParDoN(s Scope, dofn interface{}, col PCollection, opts ...Option) []PCollection

ParDoN inserts a ParDo with any number of outputs into the pipeline.

func Partition ¶ 

func Partition(s Scope, n int, fn interface{}, col PCollection) []PCollection

Partition takes a PCollection<T> and a PartitionFn, uses the PartitionFn to split the elements of the input PCollection into N partitions, and returns a []PCollection<T> that bundles N PCollection<T>s containing the split elements.

A PartitionFn has the signature `func(T) int.`

T is permitted to be a KV.

func Reshuffle ¶ 

func Reshuffle(s Scope, col PCollection) PCollection

Reshuffle copies a PCollection of the same kind and using the same element coder, and maintains the same windowing information. Importantly, it allows the result PCollection to be processed with a different sharding, in a different stage than the input PCollection.

For example, if a computation needs a lot of parallelism but produces only a small amount of output data, then the computation producing the data can run with as much parallelism as needed, while the output file is written with a smaller amount of parallelism, using the following pattern: 

pc := bigHairyComputationNeedingParallelism(scope) // PCollection<string>
resharded := beam.Reshuffle(scope, pc)                // PCollection<string>

Another use case is when one has a non-deterministic DoFn followed by one that performs externally-visible side effects. Inserting a Reshuffle between these DoFns ensures that retries of the second DoFn will always be the same, which is necessary to make side effects idempotent.

A Reshuffle will force a break in the optimized pipeline. Consequently, this operation should be used sparingly, only after determining that the pipeline without reshuffling is broken in some way and performing an extra operation is worth the cost.

func Seq ¶ 

func Seq(s Scope, col PCollection, dofns ...interface{}) PCollection

Seq is a convenience helper to chain single-input/single-output ParDos together in a sequence.

Example ¶ 
Output:

func SwapKV ¶ 

func SwapKV(s Scope, col PCollection) PCollection

SwapKV swaps the key and value for an input PCollection<KV<A,B>>. It returns a PCollection<KV<B,A>>.

func TryCoGroupByKey ¶ 

func TryCoGroupByKey(s Scope, cols ...PCollection) (PCollection, error)

TryCoGroupByKey inserts a CoGBK transform into the pipeline. Returns an error on failure.

func TryCombine ¶ 

func TryCombine(s Scope, combinefn interface{}, col PCollection, opts ...Option) (PCollection, error)

TryCombine attempts to insert a global Combine transform into the pipeline. It may fail for multiple reasons, notably that the combinefn is not valid or cannot be bound -- due to type mismatch, say -- to the incoming PCollections.

func TryCombinePerKey ¶ 

func TryCombinePerKey(s Scope, combinefn interface{}, col PCollection, opts ...Option) (PCollection, error)

TryCombinePerKey attempts to insert a per-key Combine transform into the pipeline. It may fail for multiple reasons, notably that the combinefn is not valid or cannot be bound -- due to type mismatch, say -- to the incoming PCollection.

func TryCreate ¶ 

func TryCreate(s Scope, values ...interface{}) (PCollection, error)

TryCreate inserts a fixed non-empty set of values into the pipeline. The values must be of the same type.

func TryCreateList ¶ 

func TryCreateList(s Scope, list interface{}) (PCollection, error)

TryCreateList inserts a fixed set of values into the pipeline from a slice or array. The values must be of the same type. Unlike TryCreate this supports the creation of an empty PCollection.

func TryExternal ¶ 

func TryExternal(s Scope, spec string, payload []byte, in []PCollection, out []FullType, bounded bool) ([]PCollection, error)

TryExternal attempts to perform the work of External, returning an error indicating why the operation failed.

func TryFlatten ¶ 

func TryFlatten(s Scope, cols ...PCollection) (PCollection, error)

TryFlatten merges incoming PCollections of type 'A' to a single PCollection of type 'A'. Returns an error indicating the set of PCollections that could not be flattened.

func TryGroupByKey ¶ 

func TryGroupByKey(s Scope, a PCollection) (PCollection, error)

TryGroupByKey inserts a GBK transform into the pipeline. Returns an error on failure.

func TryParDo ¶ 

func TryParDo(s Scope, dofn interface{}, col PCollection, opts ...Option) ([]PCollection, error)

TryParDo attempts to insert a ParDo transform into the pipeline. It may fail for multiple reasons, notably that the dofn is not valid or cannot be bound -- due to type mismatch, say -- to the incoming PCollections.

func TryReshuffle ¶ 

func TryReshuffle(s Scope, col PCollection) (PCollection, error)

TryReshuffle inserts a Reshuffle into the pipeline, and returns an error if the pcollection's unable to be reshuffled.

func TryWindowInto ¶ 

func TryWindowInto(s Scope, ws *window.Fn, col PCollection) (PCollection, error)

TryWindowInto attempts to insert a WindowInto transform.

func WindowInto ¶ 

func WindowInto(s Scope, ws *window.Fn, col PCollection) PCollection

WindowInto applies the windowing strategy to each element.

func (PCollection) Coder ¶ 

func (p PCollection) Coder() Coder

Coder returns the coder for the collection. The Coder is of type 'A'.

func (PCollection) IsValid ¶ 

func (p PCollection) IsValid() bool

IsValid returns true iff the PCollection is valid and part of a Pipeline. Any use of an invalid PCollection will result in a panic.

func (PCollection) SetCoder ¶ 

func (p PCollection) SetCoder(c Coder) error

SetCoder set the coder for the collection. The Coder must be of type 'A'.

func (PCollection) String ¶ 

func (p PCollection) String() string

func (PCollection) Type ¶ 

func (p PCollection) Type() FullType

Type returns the full type 'A' of the elements. 'A' must be a concrete type, such as int or KV<int,string>.

type Pipeline ¶ 

type Pipeline struct {
	// contains filtered or unexported fields
}

Pipeline manages a directed acyclic graph of primitive PTransforms, and the PCollections that the PTransforms consume and produce. Each Pipeline is self-contained and isolated from any other Pipeline. The Pipeline owns the PCollections and PTransforms and they can by used by that Pipeline only. Pipelines can safely be executed concurrently.

func NewPipeline ¶ 

func NewPipeline() *Pipeline

NewPipeline creates a new empty pipeline.

func (*Pipeline) Build ¶ 

func (p *Pipeline) Build() ([]*graph.MultiEdge, []*graph.Node, error)

Build validates the Pipeline and returns a lower-level representation for execution. It is called by runners only.

func (*Pipeline) Root ¶ 

func (p *Pipeline) Root() Scope

Root returns the root scope of the pipeline.

func (*Pipeline) String ¶ 

func (p *Pipeline) String() string

type Scope ¶ 

type Scope struct {
	// contains filtered or unexported fields
}

Scope is a hierarchical grouping for composite transforms. Scopes can be enclosed in other scopes and for a tree structure. For pipeline updates, the scope chain form a unique name. The scope chain can also be used for monitoring and visualization purposes.

func (Scope) IsValid ¶ 

func (s Scope) IsValid() bool

IsValid returns true iff the Scope is valid. Any use of an invalid Scope will result in a panic.

func (Scope) Scope ¶ 

func (s Scope) Scope(name string) Scope

Scope returns a sub-scope with the given name. The name provided may be augmented to ensure uniqueness.

func (Scope) String ¶ 

func (s Scope) String() string

type SideInput ¶ 

type SideInput struct {
	Input PCollection
}

SideInput provides a view of the given PCollection to the transformation.

Example ¶ 
Output:

type T ¶ 

type T = typex.T

T is a Universal Type used to represent "generic" types in DoFn and PCollection signatures. Each universal type is distinct from all others.

type TypeDefinition ¶ 

type TypeDefinition struct {
	// Var is the universal type defined.
	Var reflect.Type
	// T is the type it is bound to.
	T reflect.Type
}

TypeDefinition provides construction-time type information that the platform cannot infer, such as structured storage sources. These types are universal types that appear as output only. Types that are inferrable should not be conveyed via this mechanism.

type U ¶ 

type U = typex.U

U is a Universal Type used to represent "generic" types in DoFn and PCollection signatures. Each universal type is distinct from all others.

type V ¶ 

type V = typex.V

V is a Universal Type used to represent "generic" types in DoFn and PCollection signatures. Each universal type is distinct from all others.

type W ¶ 

type W = typex.W

W is a Universal Type used to represent "generic" types in DoFn and PCollection signatures. Each universal type is distinct from all others.

type Window ¶ 

type Window = typex.Window

type X ¶ 

type X = typex.X

X is a Universal Type used to represent "generic" types in DoFn and PCollection signatures. Each universal type is distinct from all others.

type Y ¶ 

type Y = typex.Y

Y is a Universal Type used to represent "generic" types in DoFn and PCollection signatures. Each universal type is distinct from all others.

type Z ¶ 

type Z = typex.Z

Z is a Universal Type used to represent "generic" types in DoFn and PCollection signatures. Each universal type is distinct from all others.

Source Files ¶

View all Source files

Directories ¶

Path Synopsis
Package artifact contains utilities for staging and retrieving artifacts.
 Package artifact contains utilities for staging and retrieving artifacts.
core
internal
io
Package log contains a re-targetable context-aware logging system.
 Package log contains a re-targetable context-aware logging system.
Package model contains the portable Beam model contracts.
 Package model contains the portable Beam model contracts.
options
Package provision contains utilities for obtaining runtime provision, information -- such as pipeline options.
 Package provision contains utilities for obtaining runtime provision, information -- such as pipeline options.
runners
testing
transforms
util
x

Jump to

Keyboard shortcuts

? : This menu
/ : Search site
f or F : Jump to
y or Y : Canonical URL
go.dev uses cookies from Google to deliver and enhance the quality of its services and to analyze traffic. Learn more.