gridsim.parallel.util
Class WorkloadFileReader

Object
  WorkloadFileReader

All Implemented Interfaces:

WorkloadModel

public class WorkloadFileReader
extends Object
implements WorkloadModel

This class is responsible for reading resource traces from a file and creating a list of jobs.

NOTE:

• This class can only take one trace file of the following format: ASCII text, zip, gz.
• If you need to load multiple trace files, then you need to create multiple instances of this class each with a unique entity name.
• If size of the trace file is huge or contains lots of traces please increase the JVM heap size accordingly by using java -Xmx option when running the simulation.
• The default job file size for sending to and receiving from a resource is Link.DEFAULT_MTU. However, you can specify the file size by using setGridletFileSize(int).
• A job run time is only for 1 PE not the total number of allocated PEs. Therefore, a Gridlet length is also calculated for 1 PE.
  For example, job #1 in the trace has a run time of 100 seconds for 2 processors. This means each processor runs job #1 for 100 seconds, if the processors have the same specification.

By default, this class follows the standard workload format as specified in http://www.cs.huji.ac.il/labs/parallel/workload/
However, you can use other format by calling the below methods before running the simulation:

• setComment(String)
• setField(int, int, int, int, int)

Since:

5.0

Author:

Anthony Sulistio and Marcos Dias de Assuncao

See Also:

Workload

Constructor Summary

WorkloadFileReader(String fileName, int rating)
Create a new WorkloadFileReader object.

Method Summary

ArrayList<WorkloadJob>	generateWorkload() Reads job information from a given file.
boolean	setComment(String cmt) Identifies the start of a comment line.
boolean	setField(int maxField, int jobNum, int submitTime, int runTime, int numProc) Tells this class what to look in the trace file.
boolean	setGridletFileSize(int size) Sets a Gridlet file size (in byte) for sending to/from a resource.

Methods inherited from class Object

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

Constructor Detail

WorkloadFileReader

public WorkloadFileReader(String fileName,
                          int rating)

Create a new WorkloadFileReader object.

Parameters:

fileName - the workload trace filename in one of the following format: ASCII text, zip, gz.

rating - the resource's PE rating

Throws:

IllegalArgumentException - This happens for the following conditions:

• the workload trace file name is null or empty
• the resource PE rating <= 0

Pre Condition:

fileName != null, rating > 0

Post Condition:

$none

Method Detail

setGridletFileSize

public boolean setGridletFileSize(int size)

Sets a Gridlet file size (in byte) for sending to/from a resource.

Parameters:

size - a Gridlet file size (in byte)

Returns:

true if it is successful, false otherwise

Pre Condition:

size > 0

Post Condition:

$none

setComment

public boolean setComment(String cmt)

Identifies the start of a comment line.

Parameters:

cmt - a character that denotes the start of a comment, e.g. ";" or "#"

Returns:

true if it is successful, false otherwise

Pre Condition:

comment != null

Post Condition:

$none

setField

public boolean setField(int maxField,
                        int jobNum,
                        int submitTime,
                        int runTime,
                        int numProc)

Tells this class what to look in the trace file. This method should be called before the start of the simulation.

By default, this class follows the standard workload format as specified in http://www.cs.huji.ac.il/labs/parallel/workload/
However, you can use other format by calling this method.

The parameters must be a positive integer number starting from 1. A special case is where jobNum == -1, meaning the job or gridlet ID starts at 1.

Parameters:

maxField - max. number of field/column in one row

jobNum - field/column number for locating the job ID

submitTime - field/column number for locating the job submit time

runTime - field/column number for locating the job run time

numProc - field/column number for locating the number of PEs required to run a job

Returns:

true if successful, false otherwise

Throws:

IllegalArgumentException - if any of the arguments are not within the acceptable ranges

Pre Condition:

maxField > 0, submitTime > 0, runTime > 0, numProc > 0

Post Condition:

$none

generateWorkload

public ArrayList<WorkloadJob> generateWorkload()

Reads job information from a given file.

Specified by:

generateWorkload in interface WorkloadModel

Returns:

the list of gridlets read from the file; null in case of failure.