com.scaleoutsoftware.soss.hserver.interop

Class ChunkedCollectionReader<KEY,VALUE,CHUNK_IDENTIFIER>

java.lang.Object

com.scaleoutsoftware.soss.hserver.interop.KeyValueProducer<KEY,VALUE>

com.scaleoutsoftware.soss.hserver.interop.ChunkedCollectionReader<KEY,VALUE,CHUNK_IDENTIFIER>

Type Parameters:

KEY - key type

VALUE - value type

CHUNK_IDENTIFIER - the type of the chunk identifier, i.e., an object which uniquely identifies a chunk in the store. Often times this is the chunk's key as stored in the data grid.

Direct Known Subclasses:

DataGridChunkedCollectionReader

public abstract class ChunkedCollectionReader<KEY,VALUE,CHUNK_IDENTIFIER>
extends KeyValueProducer<KEY,VALUE>

This is the base class for chunked collection readers. Chunked collection readers read key-value collections organized as a set of chunks, each holding a number of entries. The purpose of this class is to provide an abstraction for sequentially reading key-value entries in the same order they were put into the collection by a ChunkedCollectionWriter. To read all the objects in the chunked collection, readNext() should be called until it returns false.

The reader pre-reads a configurable number of chunks by using a thread pool to call readChunkFromTheGrid(Object, ChunkBuffer, int). There is always one chunk served at that time (i.e., its underlying buffer is used to get the next object entry). The chunks are served in the same order provided by their identifiers nextChunkIdentifier().

Constructor Summary

Constructors

Constructor and Description
ChunkedCollectionReader(com.scaleoutsoftware.soss.client.map.impl.ChunkBufferPool chunkBufferPool, int readTimeoutMilliseconds) Create the chunk reader instance.

Method Summary

Modifier and Type	Method and Description
abstract KEY	getKey() Gets the current key.
long	getMaxWaitingTime() Gets the maximum time spent waiting for unavailable chunk.
long	getTotalWaitingTime() Gets the total time spent waiting for unavailable chunks.
abstract VALUE	getValue() Gets the current value.
void	prepareChunk(CHUNK_IDENTIFIER identifier, DataInputStream dataInput, ByteBuffer buffer, int chunkIndex) This method is called once before the chunk is served.
boolean	readNext() Reads the next entry from the chunks.
void	startReading() Starts reading the chunks from the grid.

Methods inherited from class java.lang.Object

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

Constructor Detail

ChunkedCollectionReader

public ChunkedCollectionReader(com.scaleoutsoftware.soss.client.map.impl.ChunkBufferPool chunkBufferPool,
                               int readTimeoutMilliseconds)

Create the chunk reader instance.

Parameters:

chunkBufferPool - buffer pool to use for the read chunks

readTimeoutMilliseconds - time to wait for the read to complete

Method Detail

startReading

public void startReading()
                  throws IOException

Starts reading the chunks from the grid.

Throws:

IOException - if an error occurred while constructing chunk reader

readNext

public boolean readNext()
                 throws IOException,
                        TimeoutException

Reads the next entry from the chunks.

Specified by:

readNext in class KeyValueProducer<KEY,VALUE>

Returns:

true if the object was read successfully, false if there are no more objects to be read.

Throws:

IOException - if there was a ScaleOut hServer access error

TimeoutException - if the read timed out

prepareChunk

public void prepareChunk(CHUNK_IDENTIFIER identifier,
                         DataInputStream dataInput,
                         ByteBuffer buffer,
                         int chunkIndex)
                  throws IOException

This method is called once before the chunk is served. It can be used to verify the chunk header, for example.

Parameters:

identifier - Chunk identifier

dataInput - A DataInputStream view of the chunk

buffer - A ByteBuffer view of the chunk

chunkIndex - The index of the chunk in the chunk sequence

Throws:

IOException - If a reading error occurred

getKey

public abstract KEY getKey()

Gets the current key. To switch to next key-value entry, call readNext().

Specified by:

getKey in class KeyValueProducer<KEY,VALUE>

Returns:

key object

getValue

public abstract VALUE getValue()

Gets the current value. To switch to next key-value entry, call readNext().

Specified by:

getValue in class KeyValueProducer<KEY,VALUE>

Returns:

value object

getTotalWaitingTime

public long getTotalWaitingTime()

Gets the total time spent waiting for unavailable chunks. This method can be used to diagnose performance and tweak the number of threads and number of chunks to pre-read.

Returns:

time in nanoseconds

getMaxWaitingTime

public long getMaxWaitingTime()

Gets the maximum time spent waiting for unavailable chunk. This method can be used to diagnose performance and tweak the number of threads and number of chunks to pre-read.

Returns:

time in nanoseconds