Package org.roaringbitmap.bsi

Interface BitmapSliceIndex

All Known Implementing Classes:

ImmutableBitSliceIndex, MutableBitSliceIndex, RoaringBitmapSliceIndex

public interface BitmapSliceIndex

BitSliceIndex bit slice index can be used to 1. store high cardinality dim column for OLAP system. 2. high compression ratio for number Given that,we have a table T(c1,c2,c3....Cn). As we know,most database has rowId for each row. then table T is actually T(rowId,c1,c2,c3,Cn). 1. if column c1 is string, we can encode c1 using dictionary. By bsi, we can only use 33 bit slice express 2^32 cardinality dim. 2. if column c2 is int32(that is 4Byte), for 1_000_000 rows, the size of c2 is more than 3.81MB. however, by bsi, the size might be less than 1MB.

Nested Class Summary

Nested Classes

Modifier and Type	Interface	Description
static class	BitmapSliceIndex.Operation

Method Summary

Modifier and Type	Method	Description
int	bitCount()
long	getLongCardinality()
Pair<java.lang.Integer,java.lang.Boolean>	getValue(int columnId)
void	serialize(java.io.DataOutput output)
void	serialize(java.nio.ByteBuffer buffer)
int	serializedSizeInBytes()
void	setValue(int columnId, int value)	set value for bsi, setValue will set each bit slice according to the input value given that we have bsi as follow ebm: RB[1 1 1 1] slice0: RB[0 1 0 1] - slice1: RB[1 0 1 0] | slice2: RB[1 0 0 1] |--bA:bit slice Array slice3: RB[0 0 0 1] - 1 2 3 4 | ---------------- columnId or rowId for columnId 1, the value is 110 that is :6 for columnId 2, the value is 1 that is :1 for columnId 3, the value is 10 that is :2 for columnId 4, the value is 1101 that is :11
void	setValues(java.util.List<Pair<java.lang.Integer,java.lang.Integer>> values, java.lang.Integer currentMaxValue, java.lang.Integer currentMinValue)	setValues will batch set value for this bsi.

Method Detail

bitCount

int bitCount()

getLongCardinality

long getLongCardinality()

setValue

void setValue(int columnId,
              int value)

set value for bsi, setValue will set each bit slice according to the input value given that we have bsi as follow ebm: RB[1 1 1 1] slice0: RB[0 1 0 1] - slice1: RB[1 0 1 0] | slice2: RB[1 0 0 1] |--bA:bit slice Array slice3: RB[0 0 0 1] - 1 2 3 4 | ---------------- columnId or rowId for columnId 1, the value is 110 that is :6 for columnId 2, the value is 1 that is :1 for columnId 3, the value is 10 that is :2 for columnId 4, the value is 1101 that is :11

Parameters:

columnId - columnId or rowId

value - value for this columnId or rowId

getValue

Pair<java.lang.Integer,java.lang.Boolean> getValue(int columnId)

Parameters:

columnId - columnId or rowId

Returns:

the value of this columnId

setValues

void setValues(java.util.List<Pair<java.lang.Integer,java.lang.Integer>> values,
               java.lang.Integer currentMaxValue,
               java.lang.Integer currentMinValue)

setValues will batch set value for this bsi. currentMaxValue/currentMinValue are optional,it's can be compute from input value list. and avoiding bsi expend slice array capacity.

serialize

void serialize(java.nio.ByteBuffer buffer)
        throws java.io.IOException

Throws:

java.io.IOException

serialize

void serialize(java.io.DataOutput output)
        throws java.io.IOException

Throws:

java.io.IOException

serializedSizeInBytes

int serializedSizeInBytes()