org.rocksdb

枚举 IndexShorteningMode

java.lang.Object

java.lang.Enum<IndexShorteningMode>

org.rocksdb.IndexShorteningMode

所有已实现的接口:

java.io.Serializable, java.lang.Comparable<IndexShorteningMode>

public enum IndexShorteningMode
extends java.lang.Enum<IndexShorteningMode>

This enum allows trading off increased index size for improved iterator seek performance in some situations, particularly when block cache is disabled (ReadOptions.fillCache() == false and direct IO is enabled (DBOptions.useDirectReads() == true). The default mode is the best tradeoff for most use cases. This option only affects newly written tables. The index contains a key separating each pair of consecutive blocks. Let A be the highest key in one block, B the lowest key in the next block, and I the index entry separating these two blocks: [ ... A] I [B ...] I is allowed to be anywhere in [A, B). If an iterator is seeked to a key in (A, I], we'll unnecessarily read the first block, then immediately fall through to the second block. However, if I=A, this can't happen, and we'll read only the second block. In kNoShortening mode, we use I=A. In other modes, we use the shortest key in [A, B), which usually significantly reduces index size. There's a similar story for the last index entry, which is an upper bound of the highest key in the file. If it's shortened and therefore overestimated, iterator is likely to unnecessarily read the last data block from each file on each seek.

枚举常量概要

枚举常量

枚举常量和说明
kNoShortening Use full keys.
kShortenSeparators Shorten index keys between blocks, but use full key for the last index key, which is the upper bound of the whole file.
kShortenSeparatorsAndSuccessor Shorten both keys between blocks and key after last block.

方法概要

限定符和类型	方法和说明
static IndexShorteningMode	valueOf(java.lang.String name) 返回带有指定名称的该类型的枚举常量。
static IndexShorteningMode[]	values() 按照声明该枚举类型的常量的顺序, 返回 包含这些常量的数组。

从类继承的方法 java.lang.Enum

clone, compareTo, equals, finalize, getDeclaringClass, hashCode, name, ordinal, toString, valueOf

从类继承的方法 java.lang.Object

getClass, notify, notifyAll, wait, wait, wait

枚举常量详细资料

kNoShortening

public static final IndexShorteningMode kNoShortening

Use full keys.

kShortenSeparators

public static final IndexShorteningMode kShortenSeparators

Shorten index keys between blocks, but use full key for the last index key, which is the upper bound of the whole file.

kShortenSeparatorsAndSuccessor

public static final IndexShorteningMode kShortenSeparatorsAndSuccessor

Shorten both keys between blocks and key after last block.

方法详细资料

values

public static IndexShorteningMode[] values()

按照声明该枚举类型的常量的顺序, 返回 包含这些常量的数组。该方法可用于迭代 常量, 如下所示:

for (IndexShorteningMode c : IndexShorteningMode.values())
    System.out.println(c);

返回:

按照声明该枚举类型的常量的顺序返回的包含这些常量的数组

valueOf

public static IndexShorteningMode valueOf(java.lang.String name)

返回带有指定名称的该类型的枚举常量。 字符串必须与用于声明该类型的枚举常量的 标识符完全匹配。(不允许有多余 的空格字符。)

参数:

name - 要返回的枚举常量的名称。

返回:

返回带有指定名称的枚举常量

抛出:

java.lang.IllegalArgumentException - 如果该枚举类型没有带有指定名称的常量

java.lang.NullPointerException - 如果参数为空值