greenplum db 4.0 readme

About Greenplum Database 4.0.0.0 Controlled Release (CR) 1

May 15, 2010

Welcome to Greenplum Database 4.0 CR

Greenplum Database is a massively parallel processing (MPP) database server

designed to support the next generation of data warehousing and large-scale analytics

processing. It allows a cluster of servers to operate as a single database super

computer — automatically partitioning data and parallelizing queries — to achieve

CR has some known issues, which are documented in

Database 4.0 CR” on page 12. Greenplum recommends that these scenarios be

avoided or to use the appropriate work-around where applicable. Greenplum Database

release:

Greenplum Database 4.0 offers the following new features:

• Enhanced Workload Management with Dynamic Query Prioritization

• Self Healing Fault Tolerance Model with Differential Online Recovery

• Direct Dispatch Performance Optimization of Single Row Operation

• MPP Tablespace Support for Non-Uniform and SSD Segment Storage

• B-Tree and Bitmap Indexes on Column-Oriented and Append-Only Tables

• Health Monitoring Infrastructure with Email and SNMP Alerting

• Writable External Tables for Parallel Data Output

capabilities to allow database administrators (DBAs) to manage the resources

allocated to query workloads. The primary mechanism has been role-based resource

queues, which provide configurable query admission limits. By using resource queues

to set limits on incoming queries, DBAs can control the number and complexity of

active queries on the system at any given time, thereby protecting the system from

queues. In 4.0, resource queues are now always enabled.

prioritization infrastructure. Each query in the system has a priority value, which

determines the relative share of system resources provided to it. The priority of a

physical disk block level, and changes to the primary are automatically applied to the

number of improvements over prior releases:

• Automatic Failure Detection and Failover. Should a segment server become

longer a need to specify a fault action mode (read-only or continue).

• Fast Differential Recovery. Greenplum Database 4.0 keeps track of the changes

Important: Resource queues are required for all roles (users) in Greenplum Database

.

• Improved Write Performance for AO Tables. Write transactions for

and segment mirroring ensures that all modified disk blocks are synchronized to

or

statement). In queries

such as these, the query plan is not dispatched to all segments, but is targeted to the

segment that contains the affected row(s). This direct dispatch approach for this type

MPP Tablespace Support for Non-Uniform and SSD Segment

Storage

remaining tables utilize standard hard drives. This is an advanced feature for

database storage.

B-Tree and Bitmap Indexes on Column-Oriented and Append-Only

Tables

append-only storage tables, including tables using compression and/or

has the option to use an index access method rather than a full table scan when

agent,

, can be configured to run on your Greenplum master host. This agent

(

OpenView or Nagios. Greenplum Database can also be configured to send an SNMP

) to enable SNMP notifications for certain

external target. Writable external tables utilize the same Scatter-Gather Streaming

infrastructure that is used when loading data.

Object-level 'Metadata Management' Tracking and Querying

on, as well as global objects such as roles and tablespaces. This allows administrators

pg_stat_partition_operations can be used to look up actions performed on an object,