Presto 0.182 发布,Facebook 开源的大数据查询引擎

来源: 投稿
作者: 愚_者
2017-08-04 09:04:00

Presto 0.182 发布了,Presto 是 Facebook 开源的数据查询引擎,可对250PB以上的数据进行快速地交互式分析,查询的速度达到商业数据仓库的级别。据称该引擎的性能是 Hive 的 10 倍以上。

Presto 可以查询包括 Hive、Cassandra 甚至是一些商业的数据存储产品。单个 Presto 查询可合并来自多个数据源的数据进行统一分析。


General Changes

  • Fix correctness issue that causes :func:`corr` to return positive numbers for inverse correlations.

  • Fix the :doc:`/sql/explain` query plan for tables that are partitioned on TIMESTAMP or DATE columns.

  • Fix query failure when when using certain window functions that take arrays or maps as arguments (e.g., :func:`approx_percentile`).

  • Implement subtraction for all TIME and TIMESTAMP types.

  • Improve planning performance for queries that join multiple tables with a large number columns.

  • Improve the performance of joins with only non-equality conditions by using a nested loops join instead of a hash join.

  • Improve the performance of casting from JSON to ARRAY or MAP types.

  • Add a new :ref:`ipaddress_type` type to represent IP addresses.

  • Add :func:`to_milliseconds` function to convert intervals (day to second) to milliseconds.

  • Add support for column aliases in CREATE TABLE AS statements.

  • Add a config option to reject queries during cluster initialization. Queries are rejected if the active worker count is less than the query-manager.initialization-required-workers property while the coordinator has been running for less than query-manager.initialization-timeout.

  • Add :doc:`/connector/tpcds`. This connector provides a set of schemas to support the TPC Benchmark™ DS (TPC-DS).

CLI Changes

  • Fix an issue that would sometimes prevent queries from being cancelled when exiting from the pager.

Hive Changes

  • Fix reading decimal values in the optimized Parquet reader when they are backed by the int32 or int64 types.

  • Add a new experimental ORC writer implementation optimized for Presto. We have some upcoming improvements, so we recommend waiting a few releases before using this in production. The new writer can be enabled with the hive.orc.optimized-writer.enabled configuration property or with the orc_optimized_writer_enabled session property.


32 收藏
2 评论
32 收藏