ExtenDB Releases Version 1.1 of its Clustered Solution for Business Intelligence

Today ExtenDB announced the release of version 1.1 of its ExtenDB Parallel Server. The improvements to version 1.1 include more flexible data loading options, expanded PostgreSQL support, improved transaction throughput for ETL workloads, and better query planning.

As with the previous version, the Standard Edition is free and allows users to cluster together four database nodes for parallelizing queries for Business Intelligence workloads.

A more detailed list of the changes appears below:

Changes

---

• Several changes to improve ease of use of XDBLoader:

• Option -g allows loading and generation of values for serial columns.

• Option -b performs basic checks like delimiter count, not null

columns, and string length, and writes bad input lines to a target

file.

• Options -k and -y allow the load to be broken out into smaller

chunks, so that the load will continue, even if some data in one

chunk is bad.

(Note: if you are one of the few who received a preview version of

this, the -y option was formerly known as -o.)

• You can now pipe into XDBLoader.

• Added support for ROUND-ROBIN partitioning.

• Added support for the LIMIT clause.

• Added support for CLUSTER indexes.

• Added support for the TRUNCATE command.

• Added support for TEXT data type.

• Added support for the BIT and VARBIT data type.

• Added support for PostgreSQL compatible binary and hexadecimal

string notation for BLOBs.

• SHOW TABLES output is now sorted by tablename.

• Added command line history buffering to cmdline utility in Linux.

You must include the -b option to use.

• Improved statement cancelation on backends.

• Made creation of internal unique row identifier optional and is

disabled by default.

• Performance improvements for handling a larger load of concurrent

insert, update, and delete statements (hundreds per second).

• Allow UPDATE and DELETE to have uncorrelated subqueries.

• Allow JDBC connections to include "mode=P", which will

hold on to underlying connections instead of releasing them

to the pool. This is useful for ETL processes that do not rely

on bulk loading.

• Fixed a bug in xdbimpex for handling null column values.

• Fixed a bug in the Optimizer when dealing with tables that have

a large row length.

• Fixed a bug with view column naming for some views. If this affected

you, you will have to drop and recreate the view.

• Fixed a bug where not all available column statistics were being

obtained when performing ANALYZE.

This post has been migrated from a previous version of the PostgreSQL website. We apologise for any formatting issues caused by the migration.