Note: This was originally posted by an inactive account. Content was preserved by moving under an admin account.
Originally posted by: simon.wilbyHi all,
New in LAL 6.1.45 / LAL 6.0.45
Overview
Please Note: All of the Spark nodes now require Spark 1.5.0 as a minimum in the <LAE Install>/lib/java/lal/spark directory. In addition the Spark SQL Query node requires the Hive Client Configuration to be located on the laeServer machine. (See the Spark Configuration Guide for Lavastorm.pdf included in your installed documentation).
- The Archive Compress node
- The QVD File node
- The Spark SQL Query node (Experimental)
- The Spark Agg node enhancement
- The Spark Join node enhancement
- The Excel File node enhancements
Details of New Functionality
The Archive Compress node creates a compressed archive file containing one or more files. LAL-4961
The QVD File node imports data from a QVD file. LAL-3102. Additional software libraries are required in order for this node to work (commons-io-2.4.jar, irbi-fileutils-0.1.0.jar, jdbm-3.0-SNAPSHOT.jar, qvd-converter-core-1.5.0.jar and qvd-reader-1.5.0.jar).
Please contact Ralf Becher using the contact details below in order to purchase these libraries. Once you have those libraries they need to be saved to <LAE install Directory>/lib/java/lal.
IT Beratung Ralf Becher
Leipzig, Germany
ralf.becher@web.de
The Spark SQL Query node provides an interface with a Spark cluster to execute a Spark SQL query. LAL-6194
The Spark Agg node has new parameters included to allow customers to include or exclude fields from the aggregate function. LAL-5817
The Spark Join node has been enhanced to include new parameters so that users are able to filter on one or both datasets. LAL-6001
The Excel File node has a new parameter called IgnoreCellFormatting which allows the raw data underneath cell formatting to be imported to LAE. The default behavior is set to �Ignore� to allow for backwards compatibility. LAL-1554/gnats 5119
Bugfixes
The Trim Fields node now fully caters for fields with or without single quotes. LAL-5004
The Trim Fields node set to NULL if empty option has been improved to work on all fields. Also if the field names do not exist for the excluded/included option an exception is now returned. LAL-4049
Third party known issues/considerations
The Spark SQL Query node has highlighted some Apache issues with the following scenarios:
- Selecting a field with binary type fails with the exception �UnresolvedUnionException: Not in union ["bytes","null"]�. This is already raised on Apache JIRA: https://issues.apache.org/jira/browse/AVRO-1401
- Describe operation returns 3 fields (col_name, col_type, comment) however the comment field is handled as not �nullable� but returns a NULL value.
The Spark SQL Query node processes against Hive tables. When Hive tables are processed by the cluster, the minimum memory requirement is higher compared with the memory required to run other Spark nodes that do not access Hive. The DriverMemory and ExecutorMemory both have a minimum 5G threshold. Users are advised to increase this for larger environments.
New in LAL 6.1.44 / LAL 6.0.44
Overview
Please Note: All of the Spark nodes now require Spark 1.5.0 as a minimum in the <LAE Install>/lib/java/lal/spark directory (See the Spark Configuration Guide included in your installed documentation).
- The SharePoint 2010 suite of nodes enhancements
- The Spark Agg node enhancements
- The Spark Filter node enhancements
- The Spark Java Submit node enhancement
- The Spark Join node enhancement
- New Guides included
Details of New Functionality
The SharePoint 2010 suite of nodes have been enhanced to disable Forms Based Authentication. LAL-5739
The SharePoint 2010 suite of nodes have been enhanced to support situations where there is a URL-Folder Name mismatch. Please note: If there is a mismatch then additional processing is required to match up display names with underlying names used by the API. LAL-5623
The Spark Agg node has been enhanced to include the function name in the output columns. LAL-6112
The Spark Agg node now includes grouping functionality. Users are able to group aggregated data by one or more fields so that any records with identical values in the specified group fields are combined into a single record. LAL-5818
The Spark Filter node now allows a user to specify which fields are to be included or excluded on the output file(s). LAL-5814
The Spark Filter now includes the ability to filter on an additional two sets of filter criteria. LAL-5813
The Spark Java Submit node has a new parameter called SparkAssemblyJar. Users are now able to specify the JAR they wish the Spark application to execute with. Users are advised to review the �Spark Configuration Guide for Lavastorm.pdf� located in their LAE installation. LAL-6075
The Spark Join node now allows the ability to select from an additional four join types (default join type �Inner�): Left Outer, Right Outer, Full Outer, Left Semi. LAL-5998
Three new documents have been added to the install of LAL. The documents included are: �CRAN Package Install Guide for Lavastorm.pdf�, �Spark Configuration Guide for Lavastorm.pdf� and �SSL Connections Guide for Lavastorm.pdf�. LAL-6111
LAL V6.1.45 Release Notes � Lavastorm Analytics 2016 All Rights reserved Page 17
Known issues/considerations
- The Spark Sample node has been updated to return �N Records� instead of �First N Records�. When using the 'N records' option, the order of the generated sample set is not guaranteed to reflect the order of the data in the input file(s).
Attached are example graphs for the Archive Compress and Spark SQL nodes
Regards
Simon
Attachments:
LAL_Archive_Compress_Node_Example_v0.1 (1).brg
Spark SQL Query Example v1.0.brg