Class AdvancedRangeCacheImpl<T>

java.lang.Object

org.aksw.commons.io.cache.AdvancedRangeCacheImpl<T>

All Implemented Interfaces:

HasArrayOps<T>, ReadableChannelFactory<T>, ReadableChannelSource<T>

public class AdvancedRangeCacheImpl<T>
extends Object
implements ReadableChannelSource<T>

Nested Class Summary

Nested Classes

Modifier and Type	Class	Description
static class	AdvancedRangeCacheImpl.Builder<A>

Field Summary

Fields

Modifier and Type	Field	Description
protected ReadableChannelSource<T>	dataSource
protected Set<RangeRequestWorkerImpl<T>>	executors
protected ExecutorService	executorService
protected int	maxReadAheadItemCount
protected long	readBeforeSize
protected long	requestLimit
protected Slice<T>	slice
protected Duration	terminationDelay
protected int	workerBulkSize
protected ReentrantLock	workerCreationLock

Constructor Summary

Constructors

Constructor	Description
AdvancedRangeCacheImpl(ReadableChannelSource<T> dataSource, Slice<T> slice, long requestLimit, int workerBulkSize, Duration terminationDelay, int maxReadAheadItemCount, ExecutorService executorService)

Method Summary

Modifier and Type	Method	Description
ArrayOps<T>	getArrayOps()
ReadableChannelSource<T>	getDataSource()
Lock	getExecutorCreationReadLock()	A lock that when held prevents creation of workers that put data into the slice.
Set<RangeRequestWorkerImpl<T>>	getExecutors()
Slice<T>	getSlice()
static <A> AdvancedRangeCacheImpl.Builder<A>	newBuilder()
Map.Entry<RangeRequestWorkerImpl<T>, Slot<Long>>	newExecutor(long offset, long initialLength)	Creates a new worker and immediately starts it.
ReadableChannel<T>	newReadableChannel(com.google.common.collect.Range<Long> range)	Create a RequestContext for the given requestRange: (1) Claim cached pages for the start-range of the request range (2) Check the running executors for whether they are suitable for (partially) servinge the request range If so, assign tasks to those executors (3) If no running executor is suitable then add the request to the 'pending queue' If the executor service
long	size()	If the size is requested but not yet known then try to obtain it from the dataSource and if this knows it then cache it with the slice

Methods inherited from class Object

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

Methods inherited from interface ReadableChannelFactory

newReadableChannel, newReadableChannel

Methods inherited from interface ReadableChannelSource

newReadableChannel

Field Details

dataSource

protected ReadableChannelSource<T> dataSource

slice

protected Slice<T> slice

workerCreationLock

protected ReentrantLock workerCreationLock

executors

protected Set<RangeRequestWorkerImpl<T>> executors

readBeforeSize

protected long readBeforeSize

requestLimit

protected long requestLimit

terminationDelay

protected Duration terminationDelay

maxReadAheadItemCount

protected int maxReadAheadItemCount

workerBulkSize

protected int workerBulkSize

executorService

protected ExecutorService executorService

Constructor Details

AdvancedRangeCacheImpl

public AdvancedRangeCacheImpl(ReadableChannelSource<T> dataSource,
 Slice<T> slice,
 long requestLimit,
 int workerBulkSize,
 Duration terminationDelay,
 int maxReadAheadItemCount,
 ExecutorService executorService)

Method Details

getArrayOps

public ArrayOps<T> getArrayOps()

Specified by:

getArrayOps in interface HasArrayOps<T>

getDataSource

public ReadableChannelSource<T> getDataSource()

size

public long size()

If the size is requested but not yet known then try to obtain it from the dataSource and if this knows it then cache it with the slice

Specified by:

size in interface ReadableChannelSource<T>

getSlice

public Slice<T> getSlice()

getExecutors

public Set<RangeRequestWorkerImpl<T>> getExecutors()

getExecutorCreationReadLock

public Lock getExecutorCreationReadLock()

A lock that when held prevents creation of workers that put data into the slice. This allows for analyzing all existing workers during scheduling; i.e. when deciding whether for a data demand any new workers need to be created or existing ones can be reused.

newExecutor

public Map.Entry<RangeRequestWorkerImpl<T>, Slot<Long>> newExecutor(long offset,
 long initialLength)

Creates a new worker and immediately starts it. The executor creation is driven by the RangeRequestIterator which creates executors on demand whenever it detects any gaps in its read ahead range which are not served by any existing executors.

Parameters:

offset -

initialLength -

Returns:

newReadableChannel

public ReadableChannel<T> newReadableChannel(com.google.common.collect.Range<Long> range)

Create a RequestContext for the given requestRange: (1) Claim cached pages for the start-range of the request range (2) Check the running executors for whether they are suitable for (partially) servinge the request range If so, assign tasks to those executors (3) If no running executor is suitable then add the request to the 'pending queue' If the executor service

Specified by:

newReadableChannel in interface ReadableChannelFactory<T>

Parameters:

requestRange -

newBuilder

public static <A> AdvancedRangeCacheImpl.Builder<A> newBuilder()