| java.lang.Object
 org.griphyn.cPlanner.cluster.Horizontal
| Constructor Summary | |
| public | Horizontal()
The default constructor. |
| Method Summary | |
| protected void | appendAttribute(StringBuffer xmlFeed, String key, String value)
Appends an xml attribute to the xml feed. | | public String | constructID(String partitionID, int id)
Given an integer id, returns a string id that is used for the clustered
job.
Parameters:
partitionID - the id of the partition.
Parameters:
id - the integer id from which the string id has to beconstructed. | | public String | description()
Returns a textual description of the transfer implementation. | | public void | determineClusters(Partition partition)
Determine the clusters for a partition. | | public ADag | getClusteredDAG()
Returns the clustered workflow. | | public int[] | getCollapseFactor(String pool, SubInfo job, int size)
Returns the collapse factor, that is used to chunk up the jobs of a
particular type on a pool. | | public ADag | getWorkflow()
Returns a reference to the workflow that is being refined by the refiner. | | public XMLProducer | getXMLProducer()
Returns a reference to the XMLProducer, that generates the XML fragment
capturing the actions of the refiner. | | public void | initialize(ADag dag, PegasusBag bag)
| | protected void | logRefinerAction(AggregatedJob clusteredJob, JobAggregator aggregator)
Records the refiner action into the Provenace Store as a XML fragment. | | public void | parents(String partitionID, List parents)
Am empty implementation of the callout, as state is maintained
internally to determine the relations between the jobs. |
| DEFAULT_COLLAPSE_FACTOR | | final public static int DEFAULT_COLLAPSE_FACTOR(Code) | | The default collapse factor for collapsing jobs with same logical name
scheduled onto the same execution pool.
|
| DESCRIPTION | | final public static String DESCRIPTION(Code) | | A short description about the partitioner.
|
| Horizontal | | public Horizontal()(Code) | | The default constructor.
|
| appendAttribute | | protected void appendAttribute(StringBuffer xmlFeed, String key, String value)(Code) | | Appends an xml attribute to the xml feed.
Parameters:
xmlFeed - the xmlFeed to which xml is being written
Parameters:
key - the attribute key
Parameters:
value - the attribute value |
| constructID | | public String constructID(String partitionID, int id)(Code) | | Given an integer id, returns a string id that is used for the clustered
job.
Parameters:
partitionID - the id of the partition.
Parameters:
id - the integer id from which the string id has to beconstructed. The id should be unique for all theclustered jobs that are formed for a particularpartition. the id of the clustered job |
| description | | public String description()(Code) | | Returns a textual description of the transfer implementation.
a short textual description |
| determineClusters | | public void determineClusters(Partition partition) throws ClustererException(Code) | | Determine the clusters for a partition. The partition is assumed to
contain independant jobs, and multiple clusters maybe created for the
partition. Internally the jobs are grouped according to transformation name
and then according to the execution site. Each group
(having same transformation name and scheduled on same site), is then
clustered.
The number of clustered jobs created for each group is dependant on the
following VDS profiles that can be associated with the jobs.
1) bundle (dictates the number of clustered jobs that are created)
2) collapse (the number of jobs that make a single clustered job)
In case of both parameters being associated with the jobs in a group, the
bundle parameter overrides collapse parameter.
Parameters:
partition - the partition for which the clusters need to bedetermined.
throws:
ClustererException - in case of error.
See Also: VDS.BUNDLE_KEY
See Also: VDS.COLLAPSE_KEY |
| getCollapseFactor | | public int[] getCollapseFactor(String pool, SubInfo job, int size)(Code) | | Returns the collapse factor, that is used to chunk up the jobs of a
particular type on a pool. The collapse factor is determined by
getting the collapse key in the VDS namespace/profile associated with the
job in the transformation catalog. Right now tc overrides the property
from the one in the properties file that specifies per pool.
There are two orthogonal notions of bundling and collapsing. In case the
bundle key is specified, it ends up overriding the collapse key, and
the bundle value is used to generate the collapse values.
Parameters:
pool - the pool where the chunking up is occuring
Parameters:
job - the SubInfo object containing the job thatis to be chunked up together.
Parameters:
size - the number of jobs that refer to the same logicaltransformation and are scheduled on the same execution pool. int array of size 2 where int[0] is the the collapse factorint[1] is the number of jobs for whom collapsing is int[0] + 1. |
| getWorkflow | | public ADag getWorkflow()(Code) | | Returns a reference to the workflow that is being refined by the refiner.
ADAG object. |
| getXMLProducer | | public XMLProducer getXMLProducer()(Code) | | Returns a reference to the XMLProducer, that generates the XML fragment
capturing the actions of the refiner. This is used for provenace
purposes.
XMLProducer |
| initialize | | public void initialize(ADag dag, PegasusBag bag) throws ClustererException(Code) | | Initializes the Clusterer impelementation
Parameters:
dag - the workflow that is being clustered.
Parameters:
bag - the bag of objects that is useful for initialization.
throws:
ClustererException - in case of error. |
| logRefinerAction | | protected void logRefinerAction(AggregatedJob clusteredJob, JobAggregator aggregator)(Code) | | Records the refiner action into the Provenace Store as a XML fragment.
Parameters:
clusteredJob - the clustered job
Parameters:
aggregator - the aggregator that was used to create this clustered job |
| parents | | public void parents(String partitionID, List parents) throws ClustererException(Code) | | Am empty implementation of the callout, as state is maintained
internally to determine the relations between the jobs.
Parameters:
partitionID - the id of a partition.
Parameters:
parents - the list of String objects that containthe id's of the parents of the partition.
throws:
ClustererException - in case of error. |
|
|