StreamBase 5.1 Release History

The StreamBase ActivFeed Adapter was recompiled to take advantage of the latest available version of the Activ APIs.

Both StreamBase Studio and StreamBase Server are now qualified to run with Sun Java 1.6 JDKs.

Red Hat Enterprise Linux 5 and SUSE Linux Enterprise Server 10 were added to the supported platforms list. Windows Vista is now supported for running StreamBase Studio, but is not supported for deployment of StreamBase Server. Oracle 10g and MySQL 5 were added to the supported JDBC-compliant databases. See Supported Configurations for details.

This release provides a significant advance in error handling for both EventFlow and StreamSQL applications. Error conditions can now be trapped at the operator level and at the module level. Errors are reported as a tuple in a standardized error tuple format, which can be manipulated as data. You can enable an Error Output Port for each operator or adapter individually, which provides a local catch mechanism for that operator or adapter. Separately, you can add an Error Input Stream to a module or application that serves as a global catch for that module or application. You can add one or more Error Output Streams to a module or application that serves as a throw mechanism. Error tuples flow from an inner module to its containing module or application, and eventually to the StreamBase Server instance that is running the application. See Using Error Ports and Error Streams for further information.

See Using the Query Operator for details.

Pattern-matching queries can now be bounded by either elapsed time (for example, all tuples received over one day), or by a range of values in a specified field. Previously, you could use only a time-based window for pattern matching. See Detecting Patterns for details.

EventFlow applications are stored as XML files containing two distinct kinds of information: semantic elements, which control functionality, and layout elements, which control the graphical layout of items on the canvas. You can now separate semantic and layout data in EventFlow applications, and store the two types of data in separate files. You can also remove the layout data from an EventFlow application. See Using EventFlow Layout Options for details.

In previous releases, the Input Ports tab of the Module Reference Properties in EventFlows included the option Override module input schemas with incoming schemas, which strictly enforced the schema for input streams at a module reference. The option was disabled by default. This option was removed in this release, and the default condition was set as if this option was still present and enabled. Therefore, the schema associated with each input port of a Module Reference is now always passed into the referenced application, overriding any module input schema. This feature is described in Defining Module Reference Properties.

This release includes the following enhancements to the ease of use of StreamBase Studio:

In StreamBase Server configuration files, the jar and dir elements are part of the java-vm section. In StreamBase releases prior to 5.1.1, the JAR files specifed in jar elements and classes found in directories specified in dir elements were added to the system classpath. Starting with StreamBase 5.1.1, these specified JARs and classes are added to the classpath of each container. This means that loaded JAR files cannot modify the global server state, which improves the isolation of applications.

If the 5.1 behavior causes problems with a legacy application, you can override the new behavior by adding a parameter to the java-vm section named jvm-prepend-classpath. This adds the specified JARs and directories to the system classpath. For example:

<java-vm>
  <param name="jvm-prepend-classpath" value="/path1/path2/example.jar"/>
</java-vm>

StreamBase 5.0 introduced support for multiple StreamBase installations on the same Windows machine. Starting with release 5.1.1, to provide more robust support for multiple installations, when Studio starts the local instance of StreamBase Server to run or debug applications, it now automatically prepends the directory containing sbd or sbd.exe to the PATH of the Server's runtime environment. Studio also automatically populates the STREAMBASE_HOME variable in the Server's runtime environment. See StreamBase Command Prompt for important details about maintaining multiple StreamBase installations on the same machine.

In the StreamBase command line interface, you can now enter abbreviated forms of the enqueue and dequeue subcommands for sbc and jsbc. The abbreviated subcommands are enq and deq. For example, jsbc enq is equivalent to jsbc enqueue. This feature is described in sbc and sbd topics and the associated man pages.

This release adds the SBD2SBD Output Adapter, which runs within an upstream application, sending tuples to a downstream application. This adapter is used by the High Availability system, and is also available for customer use in complex StreamBase applications.

Properties were added to the SBD2SBD Input Adapter to support secure connections from upstream StreamBase Servers.

The expression language was extended with the following functions: getContainer(), getNodeName(), and systemenv(). The stdev() and stdevp() functions were enhanced to take and return a timestamp. See StreamBase Expression Language and Functions for details.

The StreamBase Java and C++ Client APIs underwent changes, primarily to support the new High Availability features. The details of the API changes are described in API Changes in 5.1.

StreamBase 5.0 introduced a new .NET API for creating StreamBase client applications, as described in Creating .NET Clients and Migrating .NET Clients. As of StreamBase 5.1.1, the StreamBase Monitor API (that is, the set of StreamBase.SB.Monitor.* classes) is now included in the new StreamBase .NET API.

When enabled, the security tagging feature allows a StreamBase application to have a security tag associated with every tuple on every stream. The security tag is automatically added by the StreamBase Server, and is not settable, although it can be accessed by means of a function call.

Security tags can be used for filtering and other computation, to allow flexible management of secure data. The security tag value on a tuple emitted from an operator is calculated based on the input tuple tag and the state of the operator. In general, stateless operators pass the tag through unmodified. Operators with state track the security level of their state, which is the aggregate security level of all tuples that contributed to their state. Operators that read state external to themselves output tags that are the maximum of the external state's tag and the input tuple's tag. Operators that modify state external to themselves modify the tag on the external state to take into account their input tuples' state.

Contact your StreamBase Systems field service representative for help configuring and using the security tagging feature.

HA support was improved for 5.1, and provides a building block approach that replaces the monolithic HA support in previous releases.

The new HA support provides: active/active server configuration, fast server restarts, and easier manual configuration of server clusters. StreamBase Servers can be grouped into flexible failover groups in which there is no requirement for coordination between servers, or for one server to assert leadership status. The StreamBase client API was modified to support HA features. A new adapter, HAHeartbeatAdapter, provides the inter-server heartbeat mechanism.

HA support in StreamBase 5.1 is an advanced feature. Contact your StreamBase Systems field service representative for help configuring and using the High Availability features.