Milestone

  • 0.2 - first release as an incubating project focused on ASF compliance
  • 0.8 - More features and more SQL compatibility
  • 0.9.0
  • 0.10.0 - Thin JDBC Driver, HBase Storage, and Better Amazon EMR support
  • 0.11.0 - Nested complex type support, In/Exist subquery support, Multi-tenancy scheduler phase 1

0.2 (released)

0.8 (release candidate)

Planned Feature Lists

ISSUE NUMDescriptionCompleted
TAJO-176Add JDBC driverO
TAJO-122Add EXPLAIN clauseO
TAJO-283Add Table PartitioningO
TAJO-16Allow Catalog to access HCatalogO
TAJO-135Upgrade Hadoop to 2.2.0O
TAJO-316Support outer joinsO
TAJO-200Support binary/text (de)serializer of RCFile, allowing Hive compatibilityO
TAJO-317Improve Tajo's Resource Manager ability, enabling disk-based and memory-based resource schedulingO
TAJO-292, TAJO-374Improve the intermediate data materialization method.V
TAJO-291Implement the push transmission type.X

0.9.0

0.10.0

0.11.0 (mid 2015)

This roadmap mainly includes significant features. Other minor features are also included in 0.11 release.

Mull-tenancy Scheduler (phase 1)

Nested Complex Type Support (phase 1)

  • Nested Record Type support (TAJO-721)
  • Array Type support (TAJO-723)
  • Flatten operator

SQL Features

Other Significant Changes

0.12.0 (2015. Q3)

(TODO)

Until 1.0.0

User Authentication and Security

  • Pluggable authentication
  • Basic user authentication
  • Kerberos support

Multi-tenancy Scheduler

  • Multi user (or queue) support
  • Fair scheduler

Yarn Support

  • Yarn cluster support

SQL features

  • Cube and Rollup
  • Scalar subquery
  • DECIMAL type

Nested Schema / Non-scalar type Support

Phase 2

Benchmarks

  • TPC-H benchmark query sets and script
  • TPC-DS benchmark query sets and script

DAG Framework

  • Parallel stage execution

Query Optimization

  • Parquet Statistics Utilization

Client Interfaces

  • ODBC Driver

Table Space Support

  • Multiple table spaces
  • Table space properties 

Generic and Pluggable Storage Handler (TAJO-337)

  • Generic storage property interface
  • JDBC-based storages (RDBMS, ..)
  • Query Optimization with considering storage properties

User-defined Function

  • CREATE/DROP FUNCTION (Runtime Function Registration/Unregistration and jars distribution)

Beyond 1.0.0

  • HDFS Cache Utilization
  • In-memory Table support
  • Push-based transmission
  • Block iteration
  • More storages like Elastic search and Cassandar
  • No labels