This document describes the development roadmap for Hypertable.  This document will change over time as we make progress and gain better visibility into user requirements.  If there is something you'd like to see on the roadmap, please let us know by posting to the Hypertable User mailing list.


  • Change SELECT semantics to match SQL.  Currently, the only cells that are returned are the ones that match in the WHERE cluase.  We plan to change the semantics to be like SQL in that the WHERE clause determines which rows are to be returned and the column list after the SELECT keyword acts as a projection on those rows.
  • Re-write query parser to produce a parse tree that can be processed with a traditional SQL query processing pipeline (re-write, optimization, etc).
  • Improve DELETE command.  The current DELETE command is limited because it can only delete columns from one row at a time.  The new DELETE command will have the following syntax:
    DELETE [column_list] FROM table_name
    WHERE boolean_expression
    If column_list is not supplied, then the cells that match in the WHERE clause will be deleted.  If column_list is supplied, then the DELETE command will behave like SELECT, the WHERE clause will select a set of rows and column_list will indicate the columns from the selected rows to be deleted.  A column_list of '*' will cause the entire row, of each selected row, to be deleted.
  • Enhance ALTER TABLE to allow for the modification of column family and access group options.
  • Add support for inter-cluster replication (log shipping)
  • Upgrade to Thrift 0.9.0


Continue to add support for SQL.  Focus will be on data analysis, including:

  • Data types
  • Aggregate functions
  • Joins