com.bigdata.btree

Class PageStats

java.lang.Object

com.bigdata.btree.BaseIndexStats

com.bigdata.btree.PageStats

Direct Known Subclasses:

BTreePageStats, HTreePageStats

public abstract class PageStats
extends BaseIndexStats

Class reports various summary statistics for nodes and leaves.

Field Summary

Fields

Modifier and Type	Field and Description
long	blobs The #of allocations that are larger than the maximum slot size in SLOT_SIZES.
long[]	histogram Histogram of the allocation slot sizes based on SLOT_SIZES.
long	leafBytes The #of bytes in the raw records for the leaves visited.
long	maxLeafBytes The min/max bytes per leaf.
long	maxNodeBytes The min/max bytes per node.
long	minLeafBytes The min/max bytes per leaf.
long	minNodeBytes The min/max bytes per node.
long	nerrors The #of errors encountered during traversal.
long	nodeBytes The #of bytes in the raw records for the nodes visited.
long	nrawRecs The number of raw record allocations and the byte size of those raw record allocations.
long	nvisited Number of nodes/leaves visited so far.
long	rawRecBytes The number of raw record allocations and the byte size of those raw record allocations.
static int[]	SLOT_SIZES This map is used to report the histogram of pages based on the actual byte count of the user data in the allocation when the backing slot size is not directly available.

Fields inherited from class com.bigdata.btree.BaseIndexStats

height, indexType, m, name, nleaves, nnodes, ntuples

Constructor Summary

Constructors

Constructor and Description
PageStats()

Method Summary

Methods

Modifier and Type	Method and Description
long	getBytesPerLeaf() The average bytes per leaf.
long	getBytesPerNode() The average bytes per node.
long	getBytesPerRawRecord() The average bytes per raw record.
String	getDataRow() Return a row of data for an index as aggregated by this PageStats object.
String	getHeaderRow() Return the header row for a table.
abstract int	getRecommendedBranchingFactor() This computes the recommended branching factor for the index based on an examination of the current branching factor, an assumed nominal page size of 8k, the min, max, and average node and leaf sizes, and the histogram of the allocation sizes for the index.
long	getTotalBytes() Return nodeBytes plus leafBytes.
String	toString()
protected void	trackSlotSize(long allocationSize) Track the histogram of allocation sizes.
void	visit(ISimpleTreeIndexAccess ndx, IAbstractNodeData node) Visit a node or leaf, updating the PageStats.

Methods inherited from class com.bigdata.btree.BaseIndexStats

writeOn

Methods inherited from class java.lang.Object

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

Field Detail

nvisited

public long nvisited

Number of nodes/leaves visited so far.

nodeBytes

public long nodeBytes

The #of bytes in the raw records for the nodes visited.

leafBytes

public long leafBytes

The #of bytes in the raw records for the leaves visited.

minNodeBytes

public long minNodeBytes

The min/max bytes per node.

maxNodeBytes

public long maxNodeBytes

The min/max bytes per node.

minLeafBytes

public long minLeafBytes

The min/max bytes per leaf.

maxLeafBytes

public long maxLeafBytes

The min/max bytes per leaf.

histogram

public final long[] histogram

Histogram of the allocation slot sizes based on SLOT_SIZES. The indices into this array are correlated with the indices into the SLOT_SIZES array. If the allocation is larger than the maximum value in SLOT_SIZES, then it is recorded in blobs instead.

blobs

public long blobs

The #of allocations that are larger than the maximum slot size in SLOT_SIZES.

nerrors

public long nerrors

The #of errors encountered during traversal.

SLOT_SIZES

public static final int[] SLOT_SIZES

This map is used to report the histogram of pages based on the actual byte count of the user data in the allocation when the backing slot size is not directly available. Allocations

nrawRecs

public long nrawRecs

The number of raw record allocations and the byte size of those raw record allocations. TODO We could also use a histogram over this information (raw records sizes).

rawRecBytes

public long rawRecBytes

The number of raw record allocations and the byte size of those raw record allocations. TODO We could also use a histogram over this information (raw records sizes).

Constructor Detail

PageStats

public PageStats()

Method Detail

trackSlotSize

protected void trackSlotSize(long allocationSize)

Track the histogram of allocation sizes.

Parameters:

allocationSize - The size of some allocation.

See Also:

histogram, blobs, SLOT_SIZES

getTotalBytes

public long getTotalBytes()

Return nodeBytes plus leafBytes.

getBytesPerNode

public long getBytesPerNode()

The average bytes per node.

getBytesPerLeaf

public long getBytesPerLeaf()

The average bytes per leaf.

getBytesPerRawRecord

public long getBytesPerRawRecord()

The average bytes per raw record.

toString

public String toString()

Overrides:

toString in class Object

getHeaderRow

public String getHeaderRow()

Return the header row for a table.

Overrides:

getHeaderRow in class BaseIndexStats

Returns:

The header row.

getDataRow

public String getDataRow()

Return a row of data for an index as aggregated by this PageStats object.

Overrides:

getDataRow in class BaseIndexStats

See Also:

getHeaderRow()

getRecommendedBranchingFactor

public abstract int getRecommendedBranchingFactor()

This computes the recommended branching factor for the index based on an examination of the current branching factor, an assumed nominal page size of 8k, the min, max, and average node and leaf sizes, and the histogram of the allocation sizes for the index.

Returns:

See Also:

Optimize RWStore allocator sizes TODO This says "branching factor", but {@link #m} is overloaded and is the branching factor for the BTree and the addressBits for the HTree. This method should be renamed (or moved into the concrete instances of this class).

visit

public void visit(ISimpleTreeIndexAccess ndx,
         IAbstractNodeData node)

Visit a node or leaf, updating the PageStats.

Note: This method MUST be extended to capture at least the initialization of the BaseIndexStats.ntuples, BaseIndexStats.nnodes, BaseIndexStats.nleaves, and BaseIndexStats.m fields.

Parameters:

ndx - The index.

node - A node or leaf in that index.