diff options
author | Patrick Wendell <pwendell@apache.org> | 2014-09-12 05:57:56 +0000 |
---|---|---|
committer | Patrick Wendell <pwendell@apache.org> | 2014-09-12 05:57:56 +0000 |
commit | 6710e746bd2713813e396a935a7a6a2bb648b5be (patch) | |
tree | b0d79b4c3399346bcd07abfed22fbde960de7c5d /releases/_posts | |
parent | ddf743b7cdfd7186610bd51606880bdce6897282 (diff) | |
download | spark-website-6710e746bd2713813e396a935a7a6a2bb648b5be.tar.gz spark-website-6710e746bd2713813e396a935a7a6a2bb648b5be.tar.bz2 spark-website-6710e746bd2713813e396a935a7a6a2bb648b5be.zip |
More changes from typo's etc for the 1.1 content.
Diffstat (limited to 'releases/_posts')
-rw-r--r-- | releases/_posts/2014-09-11-spark-release-1-1-0.md | 14 |
1 files changed, 9 insertions, 5 deletions
diff --git a/releases/_posts/2014-09-11-spark-release-1-1-0.md b/releases/_posts/2014-09-11-spark-release-1-1-0.md index 9f627b40f..621995315 100644 --- a/releases/_posts/2014-09-11-spark-release-1-1-0.md +++ b/releases/_posts/2014-09-11-spark-release-1-1-0.md @@ -13,17 +13,19 @@ meta: Spark 1.1.0 is the first minor release on the 1.X line. This release brings operational and performance improvements in Spark core along with significant extensions to Spark’s newest libraries: MLlib and Spark SQL. It also builds out Spark’s Python support and adds new components to the Spark Streaming module. Spark 1.1 represents the work of 171 contributors, the most to ever contribute to a Spark release! +To download Spark 1.1 visit the <a href="{{site.url}}downloads.html">downloads</a> page. + ### Performance and Usability Improvements -Across the board, Spark 1.1 adds features for improved stability and performance, particularly for large-scale workloads. Spark now performs [disk spilling for skewed blocks](https://issues.apache.org/jira/browse/SPARK-1777) during cache operations, guarding against memory overflows if a single RDD partition is large. Disk spilling during aggregations, introduced in Spark 1.0, has been [ported to PySpark](https://issues.apache.org/jira/browse/SPARK-2538). This release introduces a [new shuffle implementation](https://issues.apache.org/jira/browse/SPARK-2045]) optimized for very large scale shuffles. This “sort-based shuffle” will be become the default in the next release, and is now available to users. For jobs with large numbers of reducers, we recommend turning this on. This release also adds several usability improvements for monitoring the performance of long running or complex jobs. Among the changes are better [named accumulators](https://issues.apache.org/jira/browse/SPARK-2380) that display in Spark’s UI, [dynamic updating of metrics](https://issues.apache.org/jira/browse/SPARK-2099) for progress tasks, and [reporting of input metrics](https://issues.apache.org/jira/browse/SPARK-1683) for tasks that read input data. +Across the board, Spark 1.1 adds features for improved stability and performance, particularly for large-scale workloads. Spark now performs [disk spilling for skewed blocks](https://issues.apache.org/jira/browse/SPARK-1777) during cache operations, guarding against memory overflows if a single RDD partition is large. Disk spilling during aggregations, introduced in Spark 1.0, has been [ported to PySpark](https://issues.apache.org/jira/browse/SPARK-2538). This release introduces a [new shuffle implementation](https://issues.apache.org/jira/browse/SPARK-2045) optimized for very large scale shuffles. This “sort-based shuffle” will be become the default in the next release, and is now available to users. For jobs with large numbers of reducers, we recommend turning this on. This release also adds several usability improvements for monitoring the performance of long running or complex jobs. Among the changes are better [named accumulators](https://issues.apache.org/jira/browse/SPARK-2380) that display in Spark’s UI, [dynamic updating of metrics](https://issues.apache.org/jira/browse/SPARK-2099) for progress tasks, and [reporting of input metrics](https://issues.apache.org/jira/browse/SPARK-1683) for tasks that read input data. ### Spark SQL Spark SQL adds a number of new features and performance improvements in this release. A [JDBC/ODBC server](http://spark.apache.org/docs/1.1.0/sql-programming-guide.html#running-the-thrift-jdbc-server) allows users to connect to SparkSQL from many different applications and provides shared access to cached tables. A new module provides [support for loading JSON data](http://spark.apache.org/docs/1.1.0/sql-programming-guide.html#json-datasets) directly into Spark’s SchemaRDD format, including automatic schema inference. Spark SQL introduces [dynamic bytecode generation](http://spark.apache.org/docs/1.1.0/sql-programming-guide.html#other-configuration-options) in this release, a technique which significantly speeds up execution for queries that perform complex expression evaluation. This release also adds support for registering Python, Scala, and Java lambda functions as UDFs, which can then be called directly in SQL. Spark 1.1 adds a [public types API to allow users to create SchemaRDD’s from custom data sources](http://spark.apache.org/docs/1.1.0/sql-programming-guide.html#programmatically-specifying-the-schema). Finally, many optimizations have been added to the native Parquet support as well as throughout the engine. ### MLlib -MLlib adds several new algorithms and optimizations in this release. 1.1 introduces a [new library of statistical packages](https://issues.apache.org/jira/browse/SPARK-2359) which provides exploratory analytic functions. These include stratified sampling, correlations, chi-squared tests and support for creating random datasets. This release adds utilities for feature extraction ([Word2Vec](https://issues.apache.org/jira/browse/SPARK-2510) and [TF-IDF](https://issues.apache.org/jira/browse/SPARK-2511)) and feature transformation ([normalization and standard scaling](https://issues.apache.org/jira/browse/SPARK-2272)). Also new are support for [nonnegative matrix factorization](https://issues.apache.org/jira/browse/SPARK-1553) and [SVG via Lanczos](https://issues.apache.org/jira/browse/SPARK-1782). The decision tree algorithm has been added in Python and Java (https://issues.apache.org/jira/browse/SPARK-2478). A tree aggregation primitive has been added to help optimize many existing algorithms. Performance improves across the board in MLlib 1.1, with improvements of around 2-3X for many algorithms and up to 5X for large scale decision tree problems. +MLlib adds several new algorithms and optimizations in this release. 1.1 introduces a [new library of statistical packages](https://issues.apache.org/jira/browse/SPARK-2359) which provides exploratory analytic functions. These include stratified sampling, correlations, chi-squared tests and support for creating random datasets. This release adds utilities for feature extraction ([Word2Vec](https://issues.apache.org/jira/browse/SPARK-2510) and [TF-IDF](https://issues.apache.org/jira/browse/SPARK-2511)) and feature transformation ([normalization and standard scaling](https://issues.apache.org/jira/browse/SPARK-2272)). Also new are support for [nonnegative matrix factorization](https://issues.apache.org/jira/browse/SPARK-1553) and [SVG via Lanczos](https://issues.apache.org/jira/browse/SPARK-1782). The decision tree algorithm has been [added in Python and Java](https://issues.apache.org/jira/browse/SPARK-2478). A tree aggregation primitive has been added to help optimize many existing algorithms. Performance improves across the board in MLlib 1.1, with improvements of around 2-3X for many algorithms and up to 5X for large scale decision tree problems. ### GraphX and Spark Streaming -Spark streaming adds a new data source [Amazon Kinesis](https://issues.apache.org/jira/browse/SPARK-1981). For the Flume support, a new mode is support which [pulls data from Flume](https://issues.apache.org/jira/browse/SPARK-1729), simplifying deployment and providing high availability. The first of a set of [streaming machine learning algorithms](https://issues.apache.org/jira/browse/SPARK-2438) is introduced with streaming linear regression. Finally, [rate limiting](https://issues.apache.org/jira/browse/SPARK-1341) has been added for streaming inputs. GraphX adds [custom storage levels for vertices and edges](https://issues.apache.org/jira/browse/SPARK-1991) along with [improved numerical precision](https://issues.apache.org/jira/browse/SPARK-2748) across the board. Finally, GraphX adds a new label propagation algorithm. +Spark streaming adds a new data source [Amazon Kinesis](https://issues.apache.org/jira/browse/SPARK-1981). For the Apache Flume, a new mode is supported which [pulls data from Flume](https://issues.apache.org/jira/browse/SPARK-1729), simplifying deployment and providing high availability. The first of a set of [streaming machine learning algorithms](https://issues.apache.org/jira/browse/SPARK-2438) is introduced with streaming linear regression. Finally, [rate limiting](https://issues.apache.org/jira/browse/SPARK-1341) has been added for streaming inputs. GraphX adds [custom storage levels for vertices and edges](https://issues.apache.org/jira/browse/SPARK-1991) along with [improved numerical precision](https://issues.apache.org/jira/browse/SPARK-2748) across the board. Finally, GraphX adds a new label propagation algorithm. ### Other Notable Improvements - PySpark now allows [reading](https://issues.apache.org/jira/browse/SPARK-1416) and [writing](https://issues.apache.org/jira/browse/SPARK-2024) arbitrary Hadoop InputFormats, including SequenceFiles, HBase, Cassandra, Avro, and other data sources @@ -32,9 +34,10 @@ Spark streaming adds a new data source [Amazon Kinesis](https://issues.apache.or - An [overflow bug](https://issues.apache.org/jira/browse/SPARK-3190) in GraphX has been fix that affects graphs with more than 4 billion vertices ### Upgrade Notes -Spark 1.1.0 is backwards compatible with Spark 1.0.X. Some configuration options have changed which might be relevant to existing users: +Spark 1.1.0 is backwards compatible with Spark 1.0.X. Some configuration option defaults have changed which might be relevant to existing users: - * The default value of `spark.io.compression.codec` is now `snappy`. Old behavior can be restored by switching to `lzf`. + * The default value of `spark.io.compression.codec` is now `snappy` for improved memory usage. Old behavior can be restored by switching to `lzf`. + * The default value of `spark.broadcast.factory` is now `org.apache.spark.broadcast.TorrentBroadcastFactory` for improved efficiency of broadcasts. Old behavior can be restored by switching to `org.apache.spark.broadcast.HttpBroadcastFactory`. * PySpark now performs external spilling during aggregations. Old behavior can be restored by setting `spark.shuffle.spill` to `false`. * PySpark uses a new heuristic for determining the parallelism of shuffle operations. Old behavior can be restored by setting `spark.default.parallelism` to the number of cores in the cluster. @@ -188,6 +191,7 @@ Spark 1.1.0 is backwards compatible with Spark 1.0.X. Some configuration options * Ted Malaska -- Flume improvement for streaming * Teng Qiu -- bug fixes in SQL and parquet * Timothy Hunter -- bug fix in repl + * Tom Graves -- YARN support (lead) * Tor Myklebust -- ALS improvements in MLlib * U Jing -- SQL fix * Uri Laserson -- bug fix |