Apache Derby Database Engine and Embedded JDBC Driver 10.17.1.0 · org.apache.derby.impl.sql.compile.OptimizerImpl

Modifier and Type	Field and Description
private JBitSet	assignedTableMap
private CostEstimateImpl	bestCost
private int[]	bestJoinOrder
private boolean	bestJoinOrderUsedPredsFromAbove
private RowOrdering	bestRowOrdering
private CostEstimateImpl	currentCost
private OptimizerPlan	currentPlan
private RowOrdering	currentRowOrdering
private CostEstimateImpl	currentSortAvoidanceCost
private long	currentTime
private DataDictionary	dDictionary
private boolean	desiredJoinOrderFound
private CostEstimate	finalCostEstimate
private int[]	firstLookOrder
private boolean	foundABestPlan
private int	joinPosition
private JoinStrategy[]	joinStrategies
private static final int	JUMPING
private LanguageConnectionContext	lcc
private int	maxMemoryPerTable
private static final int	NO_JUMP
private JBitSet	nonCorrelatedTableMap
private boolean	noTimeout
private int	numOptimizables
private int	numTablesInQuery
private OptimizableList	optimizableList
private CostEstimateImpl	outermostCostEstimate
private OptimizerPlan	overridingPlan
private int	permuteState
private OptimizablePredicateList	predicateList
private int[]	proposedJoinOrder
private static final int	READY_TO_JUMP
private boolean	reloadBestPlan
private RequiredRowOrdering	requiredRowOrdering
private boolean	ruleBasedOptimization
private HashMap<Object, int[]>	savedJoinOrders
private CostEstimate	sortCost
private int	tableLockThreshold
private boolean	timeExceeded
private double	timeLimit
private long	timeOptimizationStarted
private boolean	useStatistics
private boolean	usingPredsPushedFromAbove
private static final int	WALK_HIGH
private static final int	WALK_LOW

Access	Constructor and Description
pack-priv	OptimizerImpl(OptimizableList optimizableList, OptimizablePredicateList predicateList, DataDictionary dDictionary, boolean ruleBasedOptimization, boolean noTimeout, boolean useStatistics, int maxMemoryPerTable, JoinStrategy[] joinStrategies, int tableLockThreshold, RequiredRowOrdering requiredRowOrdering, int numTablesInQuery, OptimizerPlan overridingPlan, LanguageConnectionContext lcc)

Method Summary

Modifier and Type	Method and Description
pack-priv void	addScopedPredicatesToList(PredicateList List of predicates to add to this OptimizerImpl's own list for pushing. pList, ContextManager cm) Add scoped predicates to this optimizer's predicateList.
public void	considerCost(Optimizable The Optimizable optimizable, OptimizablePredicateList The OptimizablePredicateList to apply predList, CostEstimate The estimated cost of the given optimizable estimatedCost, CostEstimate The cost of the tables outer to the one being optimizer - tells how many outer rows there are. outerCost) Implements org.apache.derby.iapi.sql.compile.Optimizer.considerCost. This is the version of costOptimizable for non-base-tables.
private void	costBasedCostOptimizable(Optimizable optimizable, TableDescriptor td, ConglomerateDescriptor cd, OptimizablePredicateList predList, CostEstimate outerCost) This method decides whether the given conglomerate descriptor is cheapest based on cost, rather than based on rules.
public void	costOptimizable(Optimizable The Optimizable optimizable, TableDescriptor TableDescriptor of the Optimizable td, ConglomerateDescriptor The ConglomerateDescriptor for the conglom to cost (This should change to an object to represent access paths, but for now this is OK). cd, OptimizablePredicateList The OptimizablePredicateList to apply predList, CostEstimate The cost of the tables outer to the one being optimizer - tells how many outer rows there are. outerCost) Implements org.apache.derby.iapi.sql.compile.Optimizer.costOptimizable. Cost the current Optimizable with the specified OPL.
public void	costPermutation() Implements org.apache.derby.iapi.sql.compile.Optimizer.costPermutation. Cost the current permutation.
private void	endOfRoundCleanup() Do any work that needs to be done after the current round of optimization has completed.
private CostEstimate	estimateTotalCost(OptimizablePredicateList predList, ConglomerateDescriptor cd, CostEstimate outerCost, Optimizable optimizable) Estimate the total cost of doing a join with the given optimizable.
public DataDictionary	getDataDictionary() Implements org.apache.derby.iapi.sql.compile.Optimizer.getDataDictionary. Return the DataDictionary that the Optimizer is using.
public CostEstimate	getFinalCost() Implements org.apache.derby.iapi.sql.compile.Optimizer.getFinalCost. Get the final estimated cost of the optimized query.
public JoinStrategy	getJoinStrategy(int whichStrategy) Implements org.apache.derby.iapi.sql.compile.Optimizer.getJoinStrategy. Gets a join strategy by number (zero-based).
public JoinStrategy	getJoinStrategy(String whichStrategy) Implements org.apache.derby.iapi.sql.compile.Optimizer.getJoinStrategy. Gets a join strategy by name.
public int	getLevel() Implements org.apache.derby.iapi.sql.compile.Optimizer.getLevel. Get the level of this optimizer.
public int	getMaxMemoryPerTable() Implements org.apache.derby.iapi.sql.compile.Optimizer.getMaxMemoryPerTable.
pack-priv CostEstimateImpl	getNewCostEstimate(double theCost, double theRowCount, double theSingleScanRowCount)
public boolean	getNextDecoratedPermutation() Implements org.apache.derby.iapi.sql.compile.Optimizer.getNextDecoratedPermutation. Iterate through the "decorated permutations", returning false when they are exhausted.
public boolean	getNextPermutation() Implements org.apache.derby.iapi.sql.compile.Optimizer.getNextPermutation. Iterate through the permutations, returning false when the permutations are exhausted.
public int	getNumberOfJoinStrategies() Implements org.apache.derby.iapi.sql.compile.Optimizer.getNumberOfJoinStrategies. Get the number of join strategies supported by this optimizer.
public Optimizable	getOptimizable(int idx) Implements org.apache.derby.iapi.sql.compile.Optimizer.getOptimizable. Get the ith (0-based) Optimizable being considered by this Optimizer.
public int	getOptimizableCount() Implements org.apache.derby.iapi.sql.compile.Optimizer.getOptimizableCount. Get the number of optimizables being considered by this Optimizer.
public CostEstimate	getOptimizedCost() Implements org.apache.derby.iapi.sql.compile.Optimizer.getOptimizedCost. Get the estimated cost of the optimized query
private UniqueTupleDescriptor	getTupleDescriptor(Optimizable optimizable) Get the unique tuple descriptor of the current access path for an Optimizable.
private int	initJumpState() Determine if we want to try "jumping" permutations with this OptimizerImpl, and (re-)initialize the permuteState field accordingly.
private boolean	isPushable(OptimizablePredicate pred)
pack-priv static boolean	isTableFunction(Optimizable optimizable) Return true if the optimizable is a table function
private boolean	joinOrderMeetsDependencies(int optNumber) Check to see if the optimizable corresponding to the received optNumber can legally be placed within the current join order.
public void	modifyAccessPaths() Implements org.apache.derby.iapi.sql.compile.Optimizer.modifyAccessPaths. Modify the access path for each Optimizable, as necessary.
private CostEstimate	newCostEstimate()
public void	prepForNextRound() Implements org.apache.derby.iapi.sql.compile.Optimizer.prepForNextRound. This method is called before every "round" of optimization, where we define a "round" to be the period between the last time a call to getOptimizer() (on either a ResultSetNode or an OptimizerFactory) returned _this_ OptimizerImpl and the time a call to this OptimizerImpl's getNextPermutation() method returns FALSE.
private void	pullOptimizableFromJoinOrder() Pull whatever optimizable is at joinPosition in the proposed join order from the join order, and update all corresponding state accordingly.
pack-priv void	pushPredicates(Optimizable curTable, JBitSet outerTables)
private double	recoverCostFromProposedJoinOrder(boolean sortAvoidance) Iterate through all optimizables in the current proposedJoinOrder and find the accumulated sum of their estimated costs.
private void	rememberBestCost(CostEstimate currentCost, int planType) Is the cost of this join order lower than the best one we've found so far?
private void	rewindJoinOrder()
private void	ruleBasedCostOptimizable(Optimizable optimizable, TableDescriptor td, ConglomerateDescriptor cd, OptimizablePredicateList predList, CostEstimate outerCost) This method decides whether the given conglomerate descriptor is cheapest based on rules, rather than based on cost estimates.
public void	setOuterRows(double outerRows) Implements org.apache.derby.iapi.sql.compile.Optimizer.setOuterRows. Set the estimated number of outer rows - good for optimizing nested optimizables like subqueries and join nodes.
public int	tableLockThreshold() Implements org.apache.derby.iapi.sql.compile.Optimizer.tableLockThreshold. Get the maximum number of estimated rows touched in a table before we decide to open the table with table locking (as opposed to row locking.
private OptTrace	tracer() Get the trace machinery
private boolean	tracingIsOn()
public double	uniqueJoinWithOuterTable(OptimizablePredicateList The predicate list to check predList) Implements org.apache.derby.iapi.sql.compile.Optimizer.uniqueJoinWithOuterTable. Tells whether any of the tables outer to the current one has a uniqueness condition on the given predicate list, and if so, how many times each unique key can be seen by the current table.
public void	updateBestPlanMaps(short Indicates whether to add, load, or remove the plan action, Object Object to use as the map key when adding/looking up a plan. If this is an instance of OptimizerImpl then it corresponds to an outer query; otherwise it's some Optimizable above this OptimizerImpl that could potentially reject plans chosen by this OptimizerImpl. planKey) Implements org.apache.derby.iapi.sql.compile.Optimizer.updateBestPlanMaps. Process (i.e. add, load, or remove) current best join order as the best one for some outer query or ancestor node, represented by another OptimizerImpl or an instance of FromTable, respectively.
public boolean	useStatistics() Implements org.apache.derby.iapi.sql.compile.Optimizer.useStatistics. If statistics should be considered by the optimizer while optimizing a query.

Inherited from java.lang.Object:: clone equals finalize getClass hashCode notify notifyAll toString wait wait wait

assignedTableMap	back to summary
private JBitSet assignedTableMap

bestCost	back to summary
private CostEstimateImpl bestCost

bestJoinOrder	back to summary
private int[] bestJoinOrder

bestJoinOrderUsedPredsFromAbove	back to summary
private boolean bestJoinOrderUsedPredsFromAbove

bestRowOrdering	back to summary
private RowOrdering bestRowOrdering

currentCost	back to summary
private CostEstimateImpl currentCost

currentPlan	back to summary
private OptimizerPlan currentPlan

currentRowOrdering	back to summary
private RowOrdering currentRowOrdering

currentSortAvoidanceCost	back to summary
private CostEstimateImpl currentSortAvoidanceCost

currentTime	back to summary
private long currentTime

dDictionary	back to summary
private DataDictionary dDictionary

desiredJoinOrderFound	back to summary
private boolean desiredJoinOrderFound

finalCostEstimate	back to summary
private CostEstimate finalCostEstimate

firstLookOrder	back to summary
private int[] firstLookOrder

foundABestPlan	back to summary
private boolean foundABestPlan

joinPosition	back to summary
private int joinPosition

joinStrategies	back to summary
private JoinStrategy[] joinStrategies

JUMPING	back to summary
private static final int JUMPING

lcc	back to summary
private LanguageConnectionContext lcc

maxMemoryPerTable	back to summary
private int maxMemoryPerTable

NO_JUMP	back to summary
private static final int NO_JUMP

nonCorrelatedTableMap	back to summary
private JBitSet nonCorrelatedTableMap

noTimeout	back to summary
private boolean noTimeout

numOptimizables	back to summary
private int numOptimizables

numTablesInQuery	back to summary
private int numTablesInQuery

optimizableList	back to summary
private OptimizableList optimizableList

outermostCostEstimate	back to summary
private CostEstimateImpl outermostCostEstimate

overridingPlan	back to summary
private OptimizerPlan overridingPlan

permuteState	back to summary
private int permuteState

predicateList	back to summary
private OptimizablePredicateList predicateList

proposedJoinOrder	back to summary
private int[] proposedJoinOrder

READY_TO_JUMP	back to summary
private static final int READY_TO_JUMP

reloadBestPlan	back to summary
private boolean reloadBestPlan

requiredRowOrdering	back to summary
private RequiredRowOrdering requiredRowOrdering

ruleBasedOptimization	back to summary
private boolean ruleBasedOptimization

savedJoinOrders	back to summary
private HashMap<Object, int[]> savedJoinOrders

sortCost	back to summary
private CostEstimate sortCost

tableLockThreshold	back to summary
private int tableLockThreshold

timeExceeded	back to summary
private boolean timeExceeded

timeLimit	back to summary
private double timeLimit

timeOptimizationStarted	back to summary
private long timeOptimizationStarted

useStatistics	back to summary
private boolean useStatistics

usingPredsPushedFromAbove	back to summary
private boolean usingPredsPushedFromAbove

WALK_HIGH	back to summary
private static final int WALK_HIGH

WALK_LOW	back to summary
private static final int WALK_LOW

OptimizerImpl	back to summary
pack-priv OptimizerImpl(OptimizableList optimizableList, OptimizablePredicateList predicateList, DataDictionary dDictionary, boolean ruleBasedOptimization, boolean noTimeout, boolean useStatistics, int maxMemoryPerTable, JoinStrategy[] joinStrategies, int tableLockThreshold, RequiredRowOrdering requiredRowOrdering, int numTablesInQuery, OptimizerPlan overridingPlan, LanguageConnectionContext lcc) throws StandardException

Method Detail

addScopedPredicatesToList	back to summary
pack-priv void addScopedPredicatesToList(PredicateList pList, ContextManager cm) throws StandardException Add scoped predicates to this optimizer's predicateList. This method is intended for use during the modifyAccessPath() phase of compilation, as it allows nodes (esp. SelectNodes) to add to the list of predicates available for the final "push" before code generation. Just as the constructor for this class allows a caller to specify a predicate list to use during the optimization phase, this method allows a caller to specify a predicate list to use during the modify-access-paths phase. Before adding the received predicates, this method also clears out any scoped predicates that might be sitting in OptimizerImpl's list from the last round of optimizing. This method should be in the Optimizer interface, but it relies on an argument type (PredicateList) which lives in an impl package. Parameters pList:PredicateList List of predicates to add to this OptimizerImpl's own list for pushing.

addScopedPredicatesToList

back to summary

pack-priv void addScopedPredicatesToList(PredicateList pList, ContextManager cm) throws StandardException

Add scoped predicates to this optimizer's predicateList. This method is intended for use during the modifyAccessPath() phase of compilation, as it allows nodes (esp. SelectNodes) to add to the list of predicates available for the final "push" before code generation. Just as the constructor for this class allows a caller to specify a predicate list to use during the optimization phase, this method allows a caller to specify a predicate list to use during the modify-access-paths phase. Before adding the received predicates, this method also clears out any scoped predicates that might be sitting in OptimizerImpl's list from the last round of optimizing. This method should be in the Optimizer interface, but it relies on an argument type (PredicateList) which lives in an impl package.

Parameters

pList:PredicateList: List of predicates to add to this OptimizerImpl's own list for pushing.

considerCost back to summary

public void considerCost(Optimizable optimizable, OptimizablePredicateList predList, CostEstimate estimatedCost, CostEstimate outerCost) throws StandardException

Implements org.apache.derby.iapi.sql.compile.Optimizer.considerCost.

This is the version of costOptimizable for non-base-tables.

Parameters

optimizable:Optimizable

Doc from org.apache.derby.iapi.sql.compile.Optimizer.considerCost.

The Optimizable

predList:OptimizablePredicateList

Doc from org.apache.derby.iapi.sql.compile.Optimizer.considerCost.

The OptimizablePredicateList to apply

estimatedCost:CostEstimate

Doc from org.apache.derby.iapi.sql.compile.Optimizer.considerCost.

The estimated cost of the given optimizable

outerCost:CostEstimate

Doc from org.apache.derby.iapi.sql.compile.Optimizer.considerCost.

The cost of the tables outer to the one being optimizer - tells how many outer rows there are.

Exceptions

StandardException:: Thrown on error

See Also

Optimizer#considerCost

costBasedCostOptimizable

back to summary

private void costBasedCostOptimizable(Optimizable optimizable, TableDescriptor td, ConglomerateDescriptor cd, OptimizablePredicateList predList, CostEstimate outerCost) throws StandardException

This method decides whether the given conglomerate descriptor is cheapest based on cost, rather than based on rules. It compares the cost of using the given ConglomerateDescriptor with the cost of using the best ConglomerateDescriptor so far.

costOptimizable back to summary

public void costOptimizable(Optimizable optimizable, TableDescriptor td, ConglomerateDescriptor cd, OptimizablePredicateList predList, CostEstimate outerCost) throws StandardException

Implements org.apache.derby.iapi.sql.compile.Optimizer.costOptimizable.

Doc from org.apache.derby.iapi.sql.compile.Optimizer.costOptimizable.

Cost the current Optimizable with the specified OPL. Caller is responsible for pushing all predicates which can be evaluated prior to costing.

Parameters

optimizable:Optimizable: The Optimizable
td:TableDescriptor: TableDescriptor of the Optimizable
cd:ConglomerateDescriptor: The ConglomerateDescriptor for the conglom to cost (This should change to an object to represent access paths, but for now this is OK).
predList:OptimizablePredicateList: The OptimizablePredicateList to apply
outerCost:CostEstimate: The cost of the tables outer to the one being optimizer - tells how many outer rows there are.

Exceptions

StandardException:: Thrown on error

costPermutation back to summary

public void costPermutation() throws StandardException

Implements org.apache.derby.iapi.sql.compile.Optimizer.costPermutation.

Doc from org.apache.derby.iapi.sql.compile.Optimizer.costPermutation.

Cost the current permutation. Caller is responsible for pushing all predicates which can be evaluated prior to costing.

Exceptions

StandardException:: Thrown on error

endOfRoundCleanup

back to summary

private void endOfRoundCleanup() throws StandardException

Do any work that needs to be done after the current round of optimization has completed. For now this just means walking the subtrees for each optimizable and removing the "bestPlan" that we saved (w.r.t to this OptimizerImpl) from all of the nodes. If we don't do this post-optimization cleanup we can end up consuming a huge amount of memory for deeply- nested queries, which can lead to OOM errors. DERBY-1315.

estimateTotalCost

back to summary

private CostEstimate estimateTotalCost(OptimizablePredicateList predList, ConglomerateDescriptor cd, CostEstimate outerCost, Optimizable optimizable) throws StandardException

Estimate the total cost of doing a join with the given optimizable.

Exceptions

StandardException:: Thrown on error

getDataDictionary back to summary

public DataDictionary getDataDictionary()

Implements org.apache.derby.iapi.sql.compile.Optimizer.getDataDictionary.

Doc from org.apache.derby.iapi.sql.compile.Optimizer.getDataDictionary.

Return the DataDictionary that the Optimizer is using. This is useful when an Optimizable needs to call optimize() on a child ResultSetNode.

Returns:DataDictionary: DataDictionary DataDictionary that the Optimizer is using.
See Also: org.apache.derby.iapi.sql.compile.Optimizer#getDataDictionary

getFinalCost back to summary

public CostEstimate getFinalCost()

Implements org.apache.derby.iapi.sql.compile.Optimizer.getFinalCost.

Doc from org.apache.derby.iapi.sql.compile.Optimizer.getFinalCost.

Get the final estimated cost of the optimized query. This should be the cost that corresponds to the best overall join order chosen by the optimizer, and thus this method should only be called after optimization is complete (i.e. when modifying access paths).

See Also: Sum up the cost of all of the trulyTheBestAccessPaths for the Optimizables in our list. Assumption is that we only get here after optimization has completed--i.e. while modifying access paths.

getJoinStrategy back to summary

public JoinStrategy getJoinStrategy(int whichStrategy)

Implements org.apache.derby.iapi.sql.compile.Optimizer.getJoinStrategy.

Doc from org.apache.derby.iapi.sql.compile.Optimizer.getJoinStrategy.

Gets a join strategy by number (zero-based).

See Also: Optimizer#getJoinStrategy

getJoinStrategy back to summary

public JoinStrategy getJoinStrategy(String whichStrategy)

Implements org.apache.derby.iapi.sql.compile.Optimizer.getJoinStrategy.

Doc from org.apache.derby.iapi.sql.compile.Optimizer.getJoinStrategy.

Gets a join strategy by name. Returns null if not found. The look-up is case-insensitive.

See Also: Optimizer#getJoinStrategy

getLevel back to summary

public int getLevel()

Implements org.apache.derby.iapi.sql.compile.Optimizer.getLevel.

Doc from org.apache.derby.iapi.sql.compile.Optimizer.getLevel.

Get the level of this optimizer.

Returns:int: The level of this optimizer.
See Also: Optimizer#getLevel

getMaxMemoryPerTable

back to summary

public int getMaxMemoryPerTable()

Implements org.apache.derby.iapi.sql.compile.Optimizer.getMaxMemoryPerTable.

Returns:int

Doc from org.apache.derby.iapi.sql.compile.Optimizer.getMaxMemoryPerTable.

the maximum number of bytes to be used per table.

getNewCostEstimate	back to summary
pack-priv CostEstimateImpl getNewCostEstimate(double theCost, double theRowCount, double theSingleScanRowCount)

getNextDecoratedPermutation back to summary

public boolean getNextDecoratedPermutation() throws StandardException

Implements org.apache.derby.iapi.sql.compile.Optimizer.getNextDecoratedPermutation.

Doc from org.apache.derby.iapi.sql.compile.Optimizer.getNextDecoratedPermutation.

Iterate through the "decorated permutations", returning false when they are exhausted. NOTE - Implementers are responsible for hiding tree pruning of access methods behind this method call.

Returns:boolean

boolean True - An optimizable decorated permutation remains. False - Decorated permutations are exhausted.

Exceptions

StandardException:: Thrown on error

See Also

Optimizer#getNextDecoratedPermutation

getNextPermutation back to summary

public boolean getNextPermutation() throws StandardException

Implements org.apache.derby.iapi.sql.compile.Optimizer.getNextPermutation.

Doc from org.apache.derby.iapi.sql.compile.Optimizer.getNextPermutation.

Iterate through the permutations, returning false when the permutations are exhausted. NOTE - Implementers are responsible for hiding tree pruning of permutations behind this method call.

Returns:boolean

boolean True - An optimizable permutation remains. False - Permutations are exhausted.

Exceptions

StandardException:: Thrown on error

See Also

Optimizer#getNextPermutation

getNumberOfJoinStrategies

back to summary

public int getNumberOfJoinStrategies()

Implements org.apache.derby.iapi.sql.compile.Optimizer.getNumberOfJoinStrategies.

Get the number of join strategies supported by this optimizer.

getOptimizable

back to summary

public Optimizable getOptimizable(int idx)

Implements org.apache.derby.iapi.sql.compile.Optimizer.getOptimizable.

Doc from org.apache.derby.iapi.sql.compile.Optimizer.getOptimizable.

Get the ith (0-based) Optimizable being considered by this Optimizer.

getOptimizableCount

back to summary

public int getOptimizableCount()

Implements org.apache.derby.iapi.sql.compile.Optimizer.getOptimizableCount.

Doc from org.apache.derby.iapi.sql.compile.Optimizer.getOptimizableCount.

Get the number of optimizables being considered by this Optimizer.

getOptimizedCost back to summary

public CostEstimate getOptimizedCost()

Implements org.apache.derby.iapi.sql.compile.Optimizer.getOptimizedCost.

Doc from org.apache.derby.iapi.sql.compile.Optimizer.getOptimizedCost.

Get the estimated cost of the optimized query

See Also: Optimizer#getOptimizedCost

getTupleDescriptor	back to summary
private UniqueTupleDescriptor getTupleDescriptor(Optimizable optimizable) throws StandardException Get the unique tuple descriptor of the current access path for an Optimizable.

initJumpState	back to summary
private int initJumpState() Determine if we want to try "jumping" permutations with this OptimizerImpl, and (re-)initialize the permuteState field accordingly.

isPushable	back to summary
private boolean isPushable(OptimizablePredicate pred)

isTableFunction	back to summary
pack-priv static boolean isTableFunction(Optimizable optimizable) Return true if the optimizable is a table function

joinOrderMeetsDependencies	back to summary
private boolean joinOrderMeetsDependencies(int optNumber) throws StandardException Check to see if the optimizable corresponding to the received optNumber can legally be placed within the current join order. More specifically, if the optimizable has any dependencies, check to see if those dependencies are satisified by the table map representing the current join order.

modifyAccessPaths back to summary

public void modifyAccessPaths() throws StandardException

Implements org.apache.derby.iapi.sql.compile.Optimizer.modifyAccessPaths.

Doc from org.apache.derby.iapi.sql.compile.Optimizer.modifyAccessPaths.

Modify the access path for each Optimizable, as necessary. This includes things like adding result sets to translate from index rows to base rows.

Exceptions

StandardException:: Thrown on error

See Also

Optimizer#modifyAccessPaths

newCostEstimate	back to summary
private CostEstimate newCostEstimate()

prepForNextRound

back to summary

public void prepForNextRound()

Implements org.apache.derby.iapi.sql.compile.Optimizer.prepForNextRound.

This method is called before every "round" of optimization, where we define a "round" to be the period between the last time a call to getOptimizer() (on either a ResultSetNode or an OptimizerFactory) returned _this_ OptimizerImpl and the time a call to this OptimizerImpl's getNextPermutation() method returns FALSE. Any re-initialization of state that is required before each round should be done in this method.

pullOptimizableFromJoinOrder	back to summary
private void pullOptimizableFromJoinOrder() throws StandardException Pull whatever optimizable is at joinPosition in the proposed join order from the join order, and update all corresponding state accordingly.

pushPredicates	back to summary
pack-priv void pushPredicates(Optimizable curTable, JBitSet outerTables) throws StandardException

recoverCostFromProposedJoinOrder

back to summary

private double recoverCostFromProposedJoinOrder(boolean sortAvoidance) throws StandardException

Iterate through all optimizables in the current proposedJoinOrder and find the accumulated sum of their estimated costs. This method is used to 'recover' cost estimate sums that have been lost due to the addition/subtraction of the cost estimate for the Optimizable at position "joinPosition". Ex. If the total cost for Optimizables at positions < joinPosition is 1500, and then the Optimizable at joinPosition has an estimated cost of 3.14E40, adding those two numbers effectively "loses" the 1500. When we later subtract 3.14E40 from the total cost estimate (as part of "pull" processing), we'll end up with 0 as the result--which is wrong. This method allows us to recover the "1500" that we lost in the process of adding and subtracting 3.14E40.

rememberBestCost

back to summary

private void rememberBestCost(CostEstimate currentCost, int planType) throws StandardException

Is the cost of this join order lower than the best one we've found so far? If so, remember it.

Note

If the user has specified a join order, it will be the only join order the optimizer considers, so it is OK to use costing to decide that it is the "best" join order.

Exceptions

StandardException:: Thrown on error

rewindJoinOrder	back to summary
private void rewindJoinOrder() throws StandardException

ruleBasedCostOptimizable

back to summary

private void ruleBasedCostOptimizable(Optimizable optimizable, TableDescriptor td, ConglomerateDescriptor cd, OptimizablePredicateList predList, CostEstimate outerCost) throws StandardException

This method decides whether the given conglomerate descriptor is cheapest based on rules, rather than based on cost estimates. The rules are: Covering matching indexes are preferred above all Non-covering matching indexes are next in order of preference Covering non-matching indexes are next in order of preference Heap scans are next in order of preference Non-covering, non-matching indexes are last in order of preference. In the current language architecture, there will always be a heap, so a non-covering, non-matching index scan will never be chosen. However, the optimizer may see a non-covering, non-matching index first, in which case it will choose it temporarily as the best conglomerate seen so far.

Note

This method sets the cost in the optimizable, even though it doesn't use the cost to determine which access path to choose. There are two reasons for this: the cost might be needed to determine join order, and the cost information is copied to the query plan.

setOuterRows back to summary

public void setOuterRows(double outerRows)

Implements org.apache.derby.iapi.sql.compile.Optimizer.setOuterRows.

Doc from org.apache.derby.iapi.sql.compile.Optimizer.setOuterRows.

Set the estimated number of outer rows - good for optimizing nested optimizables like subqueries and join nodes.

See Also: Optimizer#setOuterRows

tableLockThreshold back to summary

public int tableLockThreshold()

Implements org.apache.derby.iapi.sql.compile.Optimizer.tableLockThreshold.

Doc from org.apache.derby.iapi.sql.compile.Optimizer.tableLockThreshold.

Get the maximum number of estimated rows touched in a table before we decide to open the table with table locking (as opposed to row locking.

See Also: Optimizer#tableLockThreshold

tracer	back to summary
private OptTrace tracer() Get the trace machinery

tracingIsOn	back to summary
private boolean tracingIsOn()

uniqueJoinWithOuterTable back to summary

public double uniqueJoinWithOuterTable(OptimizablePredicateList predList) throws StandardException

Implements org.apache.derby.iapi.sql.compile.Optimizer.uniqueJoinWithOuterTable.

Doc from org.apache.derby.iapi.sql.compile.Optimizer.uniqueJoinWithOuterTable.

Tells whether any of the tables outer to the current one has a uniqueness condition on the given predicate list, and if so, how many times each unique key can be seen by the current table.

Parameters

predList:OptimizablePredicateList: The predicate list to check

Returns:double

<= 0 means there is no uniqueness condition > 0 means there is a uniqueness condition on an outer table, and the return value is the reciprocal of the maximum number of times the optimizer estimates that each unique key will be returned. For example, 0.5 means the optimizer thinks each distinct join key will be returned at most twice.

Exceptions

StandardException:: Thrown on error

See Also

Optimizer#uniqueJoinWithOuterTable

updateBestPlanMaps

back to summary

public void updateBestPlanMaps(short action, Object planKey) throws StandardException

Implements org.apache.derby.iapi.sql.compile.Optimizer.updateBestPlanMaps.

Process (i.e. add, load, or remove) current best join order as the best one for some outer query or ancestor node, represented by another OptimizerImpl or an instance of FromTable, respectively. Then iterate through our optimizableList and tell each Optimizable to do the same. See Optimizable.updateBestPlan() for more on why this is necessary.

Parameters

action:short: Indicates whether to add, load, or remove the plan
planKey:Object: Object to use as the map key when adding/looking up a plan. If this is an instance of OptimizerImpl then it corresponds to an outer query; otherwise it's some Optimizable above this OptimizerImpl that could potentially reject plans chosen by this OptimizerImpl.

useStatistics back to summary

public boolean useStatistics()

Implements org.apache.derby.iapi.sql.compile.Optimizer.useStatistics.

Doc from org.apache.derby.iapi.sql.compile.Optimizer.useStatistics.

If statistics should be considered by the optimizer while optimizing a query. The user may disable the use of statistics by setting the property derby.optimizer.useStatistics or by using the property useStatistics in a query.

See Also: Optimizer#useStatistics

pack-priv Class OptimizerImpl

Field Summary

Constructor Summary

Method Summary

Field Detail

Constructor Detail

Method Detail