com.netflix.hollow.core.write

Class HollowWriteStateEngine

java.lang.Object

com.netflix.hollow.core.write.HollowWriteStateEngine

All Implemented Interfaces:

HollowDataset, HollowStateEngine

public class HollowWriteStateEngine
extends java.lang.Object
implements HollowStateEngine

A HollowWriteStateEngine is our main handle to a Hollow dataset as a data producer.

A dataset changes over time. A core concept in Hollow is that the timeline for a changing dataset can be broken down into discrete data states, each of which is a complete snapshot of the data at a particular point in time. Data producers handle data states with a HollowWriteStateEngine.

A HollowWriteStateEngine cycles back and forth between two-phases:

1. Adding records
2. Writing the state

During the "adding records" phase, all of the records comprising the dataset are added to the state engine. During the "writing" phase, we can write snapshot blobs to initialize and/or delta blobs to keep up-to-date consumers of the dataset.

Each cycle between the phases will produce a state. During each cycle, all of the current records in the dataset should be re-added to the write state engine.

Constructor Summary

Constructors

Constructor and Description
HollowWriteStateEngine()
HollowWriteStateEngine(HollowObjectHashCodeFinder hasher)

Method Summary

Methods

Modifier and Type	Method and Description
int	add(java.lang.String type, HollowWriteRecord rec) Add a record to the state.
void	addAllObjectsFromPreviousCycle() Add all of the objects from the previous cycle, exactly as they were in the previous cycle.
void	addHeaderTag(java.lang.String name, java.lang.String value)
void	addHeaderTags(java.util.Map<java.lang.String,java.lang.String> headerTags)
void	addTypeState(HollowTypeWriteState writeState) Add a type to the dataset.
HollowObjectHashCodeFinder	getHashCodeFinder()
java.lang.String	getHeaderTag(java.lang.String name)
java.util.Map<java.lang.String,java.lang.String>	getHeaderTags()
long	getNextStateRandomizedTag()
java.util.List<HollowTypeWriteState>	getOrderedTypeStates()
long	getPreviousStateRandomizedTag()
HollowSchema	getSchema(java.lang.String schemaName)
java.util.List<HollowSchema>	getSchemas()
HollowTypeWriteState	getTypeState(java.lang.String typeName)
boolean	hasChangedSinceLastCycle()
boolean	isRestored()
void	overrideNextStateRandomizedTag(long nextStateRandomizedTag)
void	overridePreviousStateRandomizedTag(long previousStateRandomizedTag)
void	prepareForNextCycle() Transition from the "writing" phase of a cycle to the "adding records" phase of the next cycle.
void	prepareForWrite() Transition from the "adding records" phase of a cycle to the "writing" phase of a cycle.
void	resetToLastPrepareForNextCycle() If a state was partially constructed after the last call to prepareForNextCycle(), this call will remove all of those objects from the state engine and reset to the state it was in at the last prepareForNextCycle() call.
void	restoreFrom(HollowReadStateEngine readStateEngine) Restore from the data state contained in the provided HollowReadStateEngine.
void	setTargetMaxTypeShardSize(long targetMaxTypeShardSize) Setting a target max type shard size (specified in bytes) will limit the excess memory pool required to perform delta transitions.

Methods inherited from class java.lang.Object

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

Constructor Detail

HollowWriteStateEngine

public HollowWriteStateEngine()

HollowWriteStateEngine

public HollowWriteStateEngine(HollowObjectHashCodeFinder hasher)

Method Detail

add

public int add(java.lang.String type,
      HollowWriteRecord rec)

Add a record to the state.

addTypeState

public void addTypeState(HollowTypeWriteState writeState)

Add a type to the dataset. Should be called during the first cycle, before writing the first state.

restoreFrom

public void restoreFrom(HollowReadStateEngine readStateEngine)

Restore from the data state contained in the provided HollowReadStateEngine. This is used to continue a delta chain after a producer is restarted.

Before calling this method, the data model should be pre-initialized. This can be accomplished by:

• using the HollowWriteStateCreator
• calling HollowObjectMapper.initializeTypeState(Class) with each of the top-level classes in the data model
• adding the types via addTypeState(HollowTypeWriteState)

prepareForWrite

public void prepareForWrite()

Transition from the "adding records" phase of a cycle to the "writing" phase of a cycle.

prepareForNextCycle

public void prepareForNextCycle()

Transition from the "writing" phase of a cycle to the "adding records" phase of the next cycle.

addAllObjectsFromPreviousCycle

public void addAllObjectsFromPreviousCycle()

Add all of the objects from the previous cycle, exactly as they were in the previous cycle.

resetToLastPrepareForNextCycle

public void resetToLastPrepareForNextCycle()

If a state was partially constructed after the last call to prepareForNextCycle(), this call will remove all of those objects from the state engine and reset to the state it was in at the last prepareForNextCycle() call.

This method can be called at any time, and will leave the state engine in the same state it was in immediately after the last call to prepareForNextCycle()

hasChangedSinceLastCycle

public boolean hasChangedSinceLastCycle()

Returns:

whether or not there are differences between the current cycle and the previous cycle.

isRestored

public boolean isRestored()

getOrderedTypeStates

public java.util.List<HollowTypeWriteState> getOrderedTypeStates()

getTypeState

public HollowTypeWriteState getTypeState(java.lang.String typeName)

Returns:

the specified HollowTypeWriteState

getSchemas

public java.util.List<HollowSchema> getSchemas()

Specified by:

getSchemas in interface HollowDataset

Specified by:

getSchemas in interface HollowStateEngine

Returns:

the schemas for all types in this dataset.

getSchema

public HollowSchema getSchema(java.lang.String schemaName)

Specified by:

getSchema in interface HollowDataset

Specified by:

getSchema in interface HollowStateEngine

Returns:

the schema for the specified type in this dataset.

getHeaderTags

public java.util.Map<java.lang.String,java.lang.String> getHeaderTags()

Specified by:

getHeaderTags in interface HollowStateEngine

addHeaderTag

public void addHeaderTag(java.lang.String name,
                java.lang.String value)

addHeaderTags

public void addHeaderTags(java.util.Map<java.lang.String,java.lang.String> headerTags)

getHeaderTag

public java.lang.String getHeaderTag(java.lang.String name)

Specified by:

getHeaderTag in interface HollowStateEngine

getHashCodeFinder

public HollowObjectHashCodeFinder getHashCodeFinder()

getPreviousStateRandomizedTag

public long getPreviousStateRandomizedTag()

overridePreviousStateRandomizedTag

public void overridePreviousStateRandomizedTag(long previousStateRandomizedTag)

getNextStateRandomizedTag

public long getNextStateRandomizedTag()

overrideNextStateRandomizedTag

public void overrideNextStateRandomizedTag(long nextStateRandomizedTag)

setTargetMaxTypeShardSize

public void setTargetMaxTypeShardSize(long targetMaxTypeShardSize)

Setting a target max type shard size (specified in bytes) will limit the excess memory pool required to perform delta transitions. For use cases where all consumers are running with hollow v2.1.0 or greater, it is recommended to set this value to something reasonably small, for example 25MB. In a future release, this value will default to (25 * 1024 * 1024). It is currently set to Long.MAX_VALUE to retain backwards compatibility with pre v2.1.0 consumers.