com.scaleoutsoftware.soss.hserver.interop

Class Image<J,I,S>

java.lang.Object

com.scaleoutsoftware.soss.hserver.interop.Image<J,I,S>

All Implemented Interfaces:

Serializable

public abstract class Image<J,I,S>
extends Object
implements Serializable

This class represents an input for the specific map/reduce job which is cached in ScaleOut StateServer. The job input is a set of InputSplits, each of which is transformed into a set of key-value pairs by the record reader. The key-value objects from a single split form a "bucket" with a unique ID, which is used to refer to this set of KV-pairs. The bucket is guaranteed to be local to the specific host to minimize network I/O while replaying the job. Each image has a unique ID string used to identify the image object in the ScaleOut StateServer store. The ID string is generated based on the job configuration each time the job is submitted and is used to determine whether there is an existing image to be replayed or a new image should be created. The ID string generation algorithms are implemented by Image subclasses. When a new image is created, the input splits from the corresponding input format are wrapped with ImageInputSplits and saved in the image object. At the recording stage, the image object is updated with the IDs of the buckets used to store the chunk keys originating from the individual splits.

See Also:

Serialized Form

Constructor Summary

Constructors

Constructor and Description
Image()

Method Summary

Modifier and Type	Method and Description
abstract boolean	checkCurrent(J context, I format) Checks whether the image is current and updates it if necessary.
long	getCreationTimestamp() Retrieves the timestamp for when the image was created.
String	getImageIdString() Gets the image ID string.
List<S>	getSplits() Gets the splits.
abstract void	initializeImageIDString(J context, I format) Computes the image ID string each time the image is constructed or retrieved.

Methods inherited from class java.lang.Object

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

Constructor Detail

Image

public Image()

Method Detail

checkCurrent

public abstract boolean checkCurrent(J context,
                                     I format)
                              throws com.scaleoutsoftware.soss.client.da.StateServerException,
                                     IOException,
                                     InterruptedException

Checks whether the image is current and updates it if necessary. It returns true if the image is current and ready to be used, or false if a new image should be created to overwrite this one.

Parameters:

context - job context

format - input format

Returns:

true if the image is current

Throws:

com.scaleoutsoftware.soss.client.da.StateServerException - If there was a communication error between the JVM and ScaleOut StateServer (or if ScaleOut StateServer experienced an internal error)

IOException - If there was an error while communicating with ScaleOut StateServer

InterruptedException - If there was a class path issue

initializeImageIDString

public abstract void initializeImageIDString(J context,
                                             I format)

Computes the image ID string each time the image is constructed or retrieved. After the string is computed, it is possible that an image with this ID string already exists in the store, and we use it instead of this instance. Because of that fact, this method should initialize the image object only to the extent required to generate the ID string.

Parameters:

context - job context

format - input format

getImageIdString

public String getImageIdString()

Gets the image ID string.

Returns:

image ID string

getSplits

public List<S> getSplits()

Gets the splits.

Returns:

list of splits

getCreationTimestamp

public long getCreationTimestamp()

Retrieves the timestamp for when the image was created.

Returns:

the creation timestamp