This repository has been archived by the owner. It is now read-only.
Permalink
Branch: master
Find file Copy path
Fetching contributors…
Cannot retrieve contributors at this time
68 lines (49 sloc) 2.3 KB

DataFusion Roadmap

0.6.x - Feature parity with original POC (in progress)

I recently updated the code to use my fork of Arrow and this required considerable rework so I am currently working on restoring functionality that was previously working.

  • Upgrade to Apache Arrow 0.12.0
  • Allow query to be executed against Arrow CSV reader
  • Allow query to be executed against Arrow Parquet reader
  • Implement project push-down so that only necessary columns are loaded into memory
  • Logical query plan definition
  • SQL Parser
  • Query planner
  • Projection
  • Selection
  • Simple aggregate queries with optional GROUP BY
  • Support for MIN/MAX
  • Support for SUM
  • Support for COUNT
  • Support for COUNT(DISTINCT)
  • ORDER BY
  • Support CREATE EXTERNAL TABLE SQL to register data sources
  • SQL console and Docker image for standalone use / easy testing and benchmarking

0.7.0 - Mature query execution

The goal of this release is to support a larger percentage of real world queries and to focus on improved unit testing to ensure correctness of query execution.

  • Scalar UDFs
  • Array UDFs
  • Support nested objects with dot notation
  • JOIN support (hash join and sort merge join)
  • Better unit tests / smoke test / performance tests

0.8.0 - Parallel processing / partitions

  • Parallel execution using threads (async/await)
  • Partitioning
  • Query optimizer improvements

0.9.0 - Groundwork for distributed queries

  • Serializable logical query plan (in protobuf format)
  • Worker node that can receive and execute plan against local files
  • Write query output to local files or return results in protobuf and/or IPC format
  • Consider supporting Hive protocol to allow JDBC/ODBC clients to submit queries to a single node

1.0.0 - Distributed queries

This release will allow queries to be executed against a cluster, supporting interactive queries that return results in Arrow format or write results to disk.

  • Distributed query planner
  • Worker can delegate portions of query plan to other workers
  • Data source meta-data
  • Kubernetes support for spinning up worker nodes

1.1.0 - Usability

  • Web user interface / better tools / monitoring etc

Backlog / TBD

  • Support for S3
  • Support for HDFS
  • Authentication/authorization
  • Encryption at rest and in transit