Package org.apache.hadoop.hbase.util

Interface RegionSplitter.SplitAlgorithm

All Known Implementing Classes:
RegionSplitter.DecimalStringSplit, RegionSplitter.HexStringSplit, RegionSplitter.NumberStringSplit, RegionSplitter.UniformSplit
Enclosing class:
RegionSplitter
public static interface RegionSplitter.SplitAlgorithm
A generic interface for the RegionSplitter code to use for all it's functionality. Note that the original authors of this code use RegionSplitter.HexStringSplit to partition their table and set it as default, but provided this for your custom algorithm. To use, create a new derived class from this interface and call RegionSplitter.createPresplitTable(org.apache.hadoop.hbase.TableName, org.apache.hadoop.hbase.util.RegionSplitter.SplitAlgorithm, java.lang.String[], org.apache.hadoop.conf.Configuration) or RegionSplitter#rollingSplit(TableName, SplitAlgorithm, Configuration) with the argument splitClassName giving the name of your class.

  • Method Summary

    Modifier and Type
    Method
    Description
    byte[]
    firstRow()
    In HBase, the first row is represented by an empty byte array.
    byte[]
    lastRow()
    In HBase, the last row is represented by an empty byte array.
    String
    rowToStr(byte[] row)
    byte array representing a row in HBase
    String
    separator()
    Returns the separator character to use when storing / printing the row
    void
    setFirstRow(byte[] userInput)
    Set the first row
    void
    setFirstRow(String userInput)
    In HBase, the last row is represented by an empty byte array.
    void
    setLastRow(byte[] userInput)
    Set the last row
    void
    setLastRow(String userInput)
    In HBase, the last row is represented by an empty byte array.
    byte[]
    split(byte[] start, byte[] end)
    Split a pre-existing region into 2 regions.
    byte[][]
    split(byte[] start, byte[] end, int numSplits, boolean inclusive)
    Some MapReduce jobs may want to run multiple mappers per region, this is intended for such usecase.
    byte[][]
    split(int numRegions)
    Split an entire table.
    byte[]
    strToRow(String input)
    user or file input for row

  • Method Details

    • split

      byte[] split(byte[] start, byte[] end)
      Split a pre-existing region into 2 regions. first row (inclusive) last row (exclusive)
      Returns:
      the split row to use

    • split

      byte[][] split(int numRegions)
      Split an entire table. number of regions to split the table into user input is validated at this time. may throw a runtime exception in response to a parse failure
      Returns:
      array of split keys for the initial regions of the table. The length of the returned array should be numRegions-1.

    • split

      byte[][] split(byte[] start, byte[] end, int numSplits, boolean inclusive)
      Some MapReduce jobs may want to run multiple mappers per region, this is intended for such usecase.
      Parameters:
      start - first row (inclusive)
      end - last row (exclusive)
      numSplits - number of splits to generate
      inclusive - whether start and end are returned as split points

    • firstRow

      byte[] firstRow()
      In HBase, the first row is represented by an empty byte array. This might cause problems with your split algorithm or row printing. All your APIs will be passed firstRow() instead of empty array.
      Returns:
      your representation of your first row

    • lastRow

      byte[] lastRow()
      In HBase, the last row is represented by an empty byte array. This might cause problems with your split algorithm or row printing. All your APIs will be passed firstRow() instead of empty array.
      Returns:
      your representation of your last row

    • setFirstRow

      void setFirstRow(String userInput)
      In HBase, the last row is represented by an empty byte array. Set this value to help the split code understand how to evenly divide the first region. raw user input (may throw RuntimeException on parse failure)

    • setLastRow

      void setLastRow(String userInput)
      In HBase, the last row is represented by an empty byte array. Set this value to help the split code understand how to evenly divide the last region. Note that this last row is inclusive for all rows sharing the same prefix. raw user input (may throw RuntimeException on parse failure)

    • strToRow

      byte[] strToRow(String input)
      user or file input for row
      Returns:
      byte array representation of this row for HBase

    • rowToStr

      String rowToStr(byte[] row)
      byte array representing a row in HBase
      Returns:
      String to use for debug & file printing

    • separator

      String separator()
      Returns the separator character to use when storing / printing the row

    • setFirstRow

      void setFirstRow(byte[] userInput)
      Set the first row
      Parameters:
      userInput - byte array of the row key.

    • setLastRow

      void setLastRow(byte[] userInput)
      Set the last row
      Parameters:
      userInput - byte array of the row key.