VoltDB
6.x
VoltDB 7.0 was a major release incorporating features from previous point releases plus new capabilities. The major new features in V7.0 and later include:
Multi-cluster Cross-Datacenter Replication (XDCR) — Cross-Datacenter Replication (XCDR) supports active replication between clusters. XDCR support has been extended from just two clusters to three or more clusters. See the chapter on "Database Replication" in the Using VoltDB manual for details.
Improved configuration for database replication (DR) — The configuration of
DR clusters has been consolidated into a single configuration file element, eliminating the need for special SQL
statements and start command flags to identify the type of DR cluster being created. The <dr
role="{type}"> attribute lets you explicitly identify the type of cluster
— master, replica, or xdcr — being started.
Simplified software upgrades — The process for upgrading VoltDB software on a single cluster has been simplified, removing the need to re-initialize and manually restore the data. Now upgrades from recent versions are simply performing a shutdown --save, upgrading the software, and restarting the database. See the section on "Upgrading VoltDB Software" in the VoltDB Administrator's Guide for details.
Views on table joins — VoltDB now supports materialized views on the join of two or more tables. See the description of the CREATE VIEW statement for details.
Window functions — VoltDB now supports six window functions: RANK, DENSE_RANK, COUNT, SUM, MAX, and MIN. The window functions allow more selective calculations on the statement results than can be achieved with plain aggregation functions. See the description of the SELECT statement for details.
Increased availability and robustness — Over the past few months extensive work has been done to harden VoltDB against common issues associated with distributed systems. Efforts include working with the Jepsen tests to identify and eliminate the last few edge cases related to data reliability and redesigning the partitioning algorithms to increase the availability beyond K factor guarantees in most cases. All of these changes have occurred "under the covers", providing added reliability and availability to customers with no changes to existing applications.
Additional options for reporting and monitoring — VoltDB continues to add to the statistics and other information available from its servers. The statistics around DR have been expanded and improved to provide better visibility into the current state of the DR clusters. In particular, the @Statistics DRROLE selector provides an overview of the cluster's role and current state. VoltDb also now provides SNMP traps for a number of important cluster events, such as server outages and exceeding resource caps.
VoltDB Management Center improvements — The monitoring tab of the VoltDB Management Center has been rearranged to improve usability. Charts and graphs related to two features — export and database replication (DR) — have been broken out into tabs of their own, simplifying the Monitor tab. The SQL Query tab has also been enhanced to support multiple queries and allow the user to resize the query and results panes.
Most of the new features and capabilities in VoltDB V7.0 do not impact existing applications. However, there are a few changes that simplify and extend existing functionality that do require minor changes to the configuration when upgrading from earlier versions. Also several deprecated features have now been removed. Existing customers should take note of of the following changes:
Supported platforms
Ubuntu 12.04 is no longer a supported production platform for VoltDB. The currently supported operating systems for running production VoltDB databases are:
CentOS V6.6 or later. Including CentOS 7.0 and later
Red Hat (RHEL) V6.6 or later, including Red Hat 7.0 and later
Ubuntu 14.04 and 16.04
Changes to DR configuration
The configuration of DR clusters has improved, aggregating, the server settings into one place within the
configuration file. In the <dr> element, the role attribute identifies the role
of the cluster, obviating the need for the SET DR=ACTIVE statement in the schema and the --replica flag on the start
command. These old syntax elements are now deprecated and VoltDB V7.0 generates a warning reminding the user to use the
new syntax when they are encountered.
For users running passive DR, be sure to add the role="replica" to the
configuration file of your replica cluster. For XDCR users, add the attribute role="xdcr" as a replacement for the SET DR=ACTIVE statement.
Old export syntax removed
The old syntax for declaring and configuring export tables has been removed from the product. This means the
EXPORT TABLE statement and <export> without explicit <configuration> sub-elements no longer work. If you have not already migrated your
export configuration to the new syntax, before using VoltDB V7.0, you must:
In the schema, replace CREATE TABLE and EXPORT TABLE statements with CREATE STREAM EXPORT TO...
In the configuration file, replace a single <export> element with
<export> and one or more <configuration> sub-elements.
See the chapter on "Importing and Exporting Live Data" in the Using VoltDB manual.
Catalog mode removed
Catalog mode (compiling and loading a separate catalog JAR file) was deprecated more than two years ago. It has now been disabled in the product. It is no longer possible to compile or load a standalone catalog. Instead, please use interactive DDL for the schema and the LOAD CLASSES command in sqlcmd to load stored procedures.
Deprecated Features
The following features are deprecated as of VoltDB V7.0. Although they continue to work in V7.x, we strongly recommend users migrate to the replacement features as the deprecated features will be removed in a future major release.
SHA-1 hashing — VoltDB supports passing credentials using either SHA-1 or SHA-2 hashing. However, SHA-1 hashing is not sufficiently strong for most applications and is therefore deprecated. Use of SHA-2 hashing is recommended.
PARTITION PROCEDURE — The standalone PARTITION PROCEDURE statement is deprecated. Instead, please include the partitioning information as part of the CREATE PROCEDURE statement using the PARTITION ON clause. The combined CREATE PARTITION statement not only is more efficient, it avoids certain edge cases where procedures cannot be declared and partitioned separately as well as making the schema of the procedure more self-explanatory.
Documentation