Pentaho Software Architecture

Introduction

The purpose of this document is to provide a detailed view of the overall software components that when combined make up the entire Pentaho open source software suite as it exists today.

At a high level, the software components can be divided into a variety of forms.  In the following detailed list, the general organization includes third party libraries and components that Pentaho has needed to fork and maintain, common libraries and projects that are used in general ways, pillars that are core business analytics or data integration elements, tools that allow access to pillars, and plugins across the pillars that provide additional functionality.  These same components can be looked at from a architectural purpose point of view, including four general areas including information delivery, data management / integration, analytics / reporting, and platform services.  For each project below we categorize in both manners to give a multi-faceted view of the overall architecture of Pentaho.

Cross Cutting Architectures, Best Practices and Use Cases

This section discusses high level cross cutting software architectures and use cases.

Configuration Management

At this time, Pentaho utilizes a combination of SVN and GIT for managing the source.  Here are some related articles:

http://wiki.pentaho.com/display/PEOpen/Advanced+Git+Topics

Metadata Definitions

As we continue to build a community of projects, it's important that they share terminology and common metadata.  Here's the beginnings of capturing shared metadata to be used across all Pentaho projects:

http://wiki.pentaho.com/display/COM/Standard+MetaStore+Element+types

Javascript Development Guidelines

Pentaho's core technology is developed within the Java Platform, but more and more the need for rich browser-based applications is becoming critical.  Pentaho has a number of components that are browser only.  It's important that we share a common approach across these projects.  Here is the beginnings of our Javascript Development Guidelines:

http://wiki.pentaho.com/display/ServerDoc2x/Javascript+Development+Guidelines

Pentaho Prompting API

This is a generally useful library used in a variety of contexts, and is part of the Common UI plugin to the Pentaho Platform.

http://wiki.pentaho.com/display/ServerDoc2x/Pentaho+Prompting+API

Pentaho Coding Standards

Cross cutting coding standards for all modules of the Pentaho suite can be found on our github project, this includes configurations for the most popular IDEs.

https://github.com/pentaho/pentaho-coding-standards

Additional content needed around:

Visualizations

Logging

Plugin Architectures

Platform / BA Server Related

Scheduling and Background Execution in Pentaho User Console: /wiki/spaces/PMOPEN/pages/1249389595 

Intro for creating a REST service for the BA Server: How to create and register a new REST service from a plugin

Developing Plugins Developing Plugins

Kettle Related

Extending Kettle (Infocenter SDK)

UI Technologies

Datasources

Detailed Software Listing

This detailed software listing is organized in the general order in which software components are dependent on one another, although it should not be used as the official build order of Pentaho.

Third Party Maintained Forks
Common Components
Pillars
Tools
Plugins

Third Party Maintained Forks

It is Pentaho's intention to avoid having to fork and maintain third party open source software, but on a few occasions it has been necessary.  The following list is of the current third party maintained forks that Pentaho includes in our product.

kettle-vfs

Kettle VFS is a maintained fork of Apache Commons VFS

Source Path: svn://source.pentaho.org/svnkettleroot/kettle-vfs

Architectural Owner: Matt Casters

Architectural Area: Data Management / Integration

hive

Due to the dynamic nature of Hadoop, Pentaho currently maintains our own Hive JDBC Driver implementation

Source Path: https://github.com/pentaho/hive

Architectural Owner: Will Gorman

Architectural Area: Data Management / Integration

pentaho-ofc4j

Pentaho ChartBeans Flash components, which are still used by Pentaho Dashboards and Action Sequences, are based on Open Flash Chart.  OFC4J is a Java to JSON converter that is used to generate the correct metadata for the charts on the server that is no longer maintained by the creator of the project.

Source Path: https://github.com/pentaho/pentaho-ofc4j

Architectural Owner: Will Gorman

Architectural Area: Information Delivery

Common Components

This is a list of all the common libraries that Pentaho maintains that are included as part of the Pentaho Suite of technologies.   Each common component has a specific purpose, and may be used by one or more pillars.

subfloor

Subfloor is Pentaho's common build system, based on ant and used by all projects for compilation, assembly, unit testing and code coverage.

Source Path: https://code.google.com/p/subfloor/  (Note that this location is out of date and should be transitioned to GitHub)

Architectural Owner: Will Gorman

Architectural Area: Build

pentaho-commons-database

This commons project is a GWT thin client of the shared database dialog.  The submodule pentaho-database-model was an attempt at a thin Kettle DatabaseMeta implementation, which includes a dialect and JDBC Metadata architecture.

Source Path: https://github.com/pentaho/pentaho-commons-database

Architectural Owner: Will Gorman

Architectural Area: Data Management / Integration

pentaho-connections

This commons project provides an API for interacting with platform connections, usually within the context of xactions.  Pentaho Metadata also uses this API for providing access to metadata in result sets.

Source Path: https://github.com/pentaho/pentaho-connections

Architectural Owner: Will Gorman

Architectural Area: Platform Services

pentaho-hdfs-vfs

This commons project provides an apache-vfs implementation of Hadoop HDFS.  This is used by Kettle for accessing HDFS in a number of contexts.

Source Path: https://github.com/pentaho/pentaho-hdfs-vfs

Architectural Owner: Will Gorman

Architectural Area: Data Management / Integration

pentaho-vfs

This is a single xml file project that allows the loading of the platform's vfs driver, that maps to the scheme "solution:"

Source Path: svn://source.pentaho.org/svnroot/pentaho-commons/pentaho-vfs

Architectural Owner: Will Gorman

Architectural Area: Platform Services

pentaho-versionchecker

This project hosts the APIs and basic logic for checking the version information of a currently running platform.

Source Path: https://github.com/pentaho/pentaho-versionchecker

Architectural Owner: Will Gorman

Architectural Area: Platform Services

pentaho-palo-core

This project is an abstraction layer that insulates the uses of Palo within the product so there is no compile time dependency, due to Palo's GPL licensing.

Source Path: https://github.com/pentaho/pdi-palo-core

Architectural Owner: Matt Casters

Architectural Area: Data Management / Integration

pentaho-registry

The Pentaho Registry project was developed in the context of Instaview to provide an API for a system wide metadata index for purposes of understanding relationships and lineage.

Source Path: https://github.com/pentaho/pentaho-registry

Architectural Owner: Will Gorman

Architectural Area: Platform Services

metastore

The Pentaho Metastore is an API that defines a simple way for interacting with a global metadata repository, and will be used as the main API for managing and storing global metadata definitions, such as connections, etc.

Source Path: https://github.com/pentaho/metastore

Architectural Owner: Matt Casters

Architectural Area: Platform Services

pentaho-cwm (deprecated)

This is a legacy project that defines the Pentaho to CWM object model API mapping, the code in the project is mostly generated, and is deprecated with the new Pentaho Metadata thin API.  This module can be completely removed once Pentaho Metadata Editor is replaced by a new editor.

Source Path: https://github.com/pentaho/pentaho-cwm

Architectural Owner: Will Gorman

Architectural Area: Analytics / Reporting (Pentaho Metadata)

pentaho-metadata

The Pentaho Metadata project defines a reporting metadata architecture and implementations.  The core components of Pentaho Metadata API are the Business Query (MQL), Logical Model, and Physical Models (CSV and SQL).  Long term plans are to merge these concepts into Pentaho Mondrian and the Metastore, but this will be an evolutionary approach and will take time before Pentaho Metadata is deprecated.

Source Path: https://github.com/pentaho/pentaho-metadata

Architectural Owner: Will Gorman

Architectural Area: Analytics / Reporting

pentaho-commons-gwt-modules

The gwt-modules project consists of the gwt-widgets and gwt-widgets-samples submodules.  This is a common area to keep reused gwt capabilities for our downstream GWT-based projects, including at this time pentaho user console, data source wizard, and others.

Source Path: https://github.com/pentaho/pentaho-commons-gwt-modules

Architectural Owner: Nick Baker

Architectural Area: Information Delivery

pentaho-xul

This is a general UI framework developed by Pentaho to address our cross platform UI requirements, it includes an implementation for Swing, GWT and SWT.  One of the key requirements to this framework is pluggability supported via overlays.  Major projects using this framework include the Aggregation Designer, the Modeler, PUC, Report Designer, Spoon, MQL Editor, and others.

Source Path: https://github.com/pentaho/pentaho-commons-xul

Architectural Owner: Nick Baker

Architectural Area: Information Delivery

modeler

This project provides an easy to use Pentaho Modeler, which is able to generate both Pentaho Metadata and Mondrian file formats.  It was originally built within the context of Agile BI, but is also currently used in the Data Source Wizard.  The plan is to eventually replace Mondrian Schema Workbench and Pentaho Metadata Editor with this UI.

Source Path: https://github.com/pentaho/modeler

Architectural Owner: Nick Baker

Architectural Area: Analytics / Reporting

pentaho-mql-editor

This project provides a simple UI for business users to define Pentaho Metadata Queries.  It is based on XUL so it can be used in Swing, SWT and GWT.

Source Path: https://github.com/pentaho/mql-editor

Architectural Owner: Nick Baker

Architectural Area: Information Delivery

pentaho-actionsequence-dom

This project provides a document object model for core action sequence file parsing.  It also manages validation of action sequence files (.xaction).  In the long term Pentaho is replacing Action Sequence functionality with Kettle Transformations and Job capabilities.

Source Path: https://github.com/pentaho/pentaho-actionsequence-dom

Architectural Owner: Will Gorman

Architectural Area: Platform Services

pentaho-chartbeans (deprecated)

This project was Pentaho's early attempt at a common Chart definition and API.  We have sense began a transition to the Visualization API, which is part of the pentaho platform plugin common-ui.  This project is still in use today by Dashboard Designer and Action Sequences, but will be phased out in a future release.

Source Path: https://github.com/pentaho/pentaho-chartbeans

Architectural Owner: Will Gorman

Architectural Area: Information Delivery

Pillars

These projects make up the core of Pentaho's architecture, and are referred to as the BI Pillars within Pentaho.

mondrian

Mondrian is a relational OLAP engine.  It supports Multidimensional Expressions (MDX) and requires a JDBC based star schema.  It includes a Swing based schema designer called Schema Workbench.

Source Location: https://github.com/pentaho/mondrian

Architectural Owner: Julian Hyde

Architectural Area: Analytics / Reporting

pentaho-reporting

Pentaho Reporting is a banded reporting engine that supports many different datasources (JDBC, Kettle, etc) and outputs (PDF, Excel, CSV, etc).  It includes a Swing based designer called Report Designer.

Source Location: https://github.com/pentaho/pentaho-reporting

Architectural Owner: Thomas Morgner

Architectural Area: Analytics / Reporting

kettle

Kettle is a metadata-based data integration engine.  There are two runtimes, data transformation and job orchestration.  Kettle includes an SWT based designer called Spoon for design and execution, as well as a server called Carte for remote execution and command line tools for execution as well.

Source Location: https://github.com/pentaho/pentaho-kettle

Architectural Owner: Matt Casters

Architectural Area: Data Management / Integration

pentaho-platform

The Pentaho Platform, which is sometimes synonymous with the Business Analytics Server, is primarily a server runtime and user interface (referred to as Pentaho User Console) for hosting Business Analytics Applications, such as report execution, OLAP, dashboards, and transformation execution.  The platform contains a legacy workflow engine called Action Sequences, that are slowly being phased out and replaced with Kettle Jobs and Transformations.

Source Location: https://github.com/pentaho/pentaho-platform

Architectural Owner: Nick Baker

Architectural Area: Platform Services

Related Architectural Documentation:

weka

Weka is a suite of machine learning software written in Java.

Source Location: https://svn.cms.waikato.ac.nz/svn/weka/

Architectural Owner: Mark Hall

Architectural Area: Analytics / Reporting

Tools

Many of the applications that make up Pentaho are included within the pillars mentioned above, these remaining tools are not managed within a pillar and thus referenced separately.

pentaho-agg-designer

The Aggregation Designer is a wizard that allows for manual and automatic creation of aggregates for use with Mondrian.  This application is based on pentaho-xul, and by default runs within the Swing container. This functionality will eventually be merged into the Pentaho Modeler.

Source Location: https://github.com/pentaho/pentaho-aggdesigner

Architectural Owner: Will Gorman

Architectural Areas: Analytics / Reporting

pentaho-metadata-editor

The Pentaho Metadata Editor allows for editing of Pentaho Metadata schemas.  This application was developed using SWT.  This functionality will eventually be merged into the Pentaho Modeler.

Source Location: https://github.com/pentaho/pentaho-metadata-editor

Architectural Owner: Will Gorman

Architectural Areas: Analytics / Reporting

Plugins

Many of the pillars have pluggable entry points that allow for extensibility of the suite.  This is a listing of the plugins that are bundled by default within the Pentaho community suite.

Pentaho Platform Plugins

common-ui

common-ui hosts a number of shared thin client and server components used by other plugins.  Major common components within common-ui include the thin client data api, the visualization API, a shared client-side prompting library, general common ui components, and more.

Source Location: https://github.com/pentaho/pentaho-platform-plugin-common-ui

Architectural Owner: Nick Baker

Architectural Area: Information Delivery

data-access

data-access is a plugin that hosts the data source wizard and other common data access components such as the database connection dialog and the thin modeler.  The client code is based on GWT-XUL, and the back-end web services are a mix of GWT-RPC and SOAP based web service calls.

The datasource wizard allows for creation of Mondrian and Pentaho Metadata based datasources.  There is a CSV option that stages the data into a single table, and a database option where users can specify relationships between tables.  The datasource wizard will automatically generate schemas that then can be customized later by the user.

Source Location: https://github.com/pentaho/data-access

Architectural Owner: Nick Baker

Architectural Area: Data Management / Integration

cdf

cdf is short for "community dashboard framework", which provides a general framework for dashboarding, including a pluggable component architecture.  WebDetails primarily drives the direction of CDF.

Source Location: https://github.com/webdetails/cdf

Project Homepage: http://www.webdetails.pt/ctools/cdf.html

Architectural Owner: Pedro Alves

Architectural Area: Information Delivery

cda (not bundled)

Community Data Access (CDA) is a Pentaho plugin designed for accessing data with great flexibility.  It is used in a number of contexts within the community, and also used as a library in Pentaho Dashboard Designer.  CDA utilizes Pentaho Reporting's Data Factory and Kettle's Transformation Engines for data access.

Source Location: https://github.com/webdetails/cda

Project Homepage: http://www.webdetails.pt/ctools/cda.html

Architectural Owner: Pedro Alves

Architectural Area: Data Management / Integration

reporting

The reporting plugin contains the elements necessary for executing Pentaho Reporting within the context of the platform.  Included is a thin- client report viewer based on CDF, and backend execution components for running and scheduling reports.

Source Location: https://github.com/pentaho/pentaho-platform-plugin-reporting

Architectural Owner: Thomas Morgner

Architectural Area: Information Delivery

Architectural overview: DOC and PPT

pdi

The PDI plugin allows native execution of transformations and jobs within the BI Server.  At this time it is not bundled by default in the BI Server.

Source Location: https://github.com/pentaho/pdi-platform-plugin

Architectural Owner: Nick Baker

Architectural Area: Data Management / Integration

Kettle Plugins

agile-bi

The Agile BI Plugin allows a Spoon user to quickly visualize a relational database table in a free embedded version of Pentaho Analyzer.  Agile BI will automatically generate a mondrian schema, It includes the pentaho-modeler for editing the schema after auto-creation.

Source Location: https://github.com/pentaho/pdi-agile-bi-plugin

Archtiectural Owner: Nick Baker

Architectural Area: Information Delivery

big-data-plugin

The Big Data Plugin contains all the related big data integration components, including HDFS, Map Reduce, Visual MapReduce, Pig, Oozie, Sqoop, Hive, Avro, HBase and CouchDB components.

Source Location: https://github.com/pentaho/big-data-plugin

Architectural Owner: Matt Casters

Architectural Area: Data Management / Integration

pentaho-hadoop-shims

This Hadoop Shims library is a sub-component of the big-data-plugin and contains the necessary adaptive layers between the big data kettle integration components and specific hadoop versions.

Source Location: https://github.com/pentaho/pentaho-hadoop-shims

Architectural Owner: Matt Casters

Architectural Area: Data Management / Integration

pentaho-mongodb-plugin

This plugin includes input and output steps for MongoDB.

Source Location: https://github.com/pentaho/pentaho-mongodb-plugin

Architectural Owner: Mark Hall

Architectural Area: Data Management / Integration

pentaho-cassandra-plugin

This plugin includes input and output steps for Cassandra

Source Location: https://github.com/pentaho/pentaho-cassandra-plugin

Architectural Owner: Mark Hall

Architectural Area: Data Management / Integration

pentaho-vertica-bulkloader

This plugin includes an HP Vertica bulk loader.

Source Location: https://github.com/pentaho/pentaho-vertica-bulkloader

Architectural Owner: Matt Casters

Architectural Area: Data Management / Integration

ArffOutput

The ARFF output plugin is a tool that allows you to output data from Kettle to a file in WEKA's Attribute Relation File Format (ARFF).

Source Location: https://github.com/pentaho/pdi-weka-arff-output-plugin

Architectural Owner: Mark Hall

Architectural Area: Analytics / Reporting

Documentation Link: http://wiki.pentaho.com/display/DATAMINING/Using+the+ARFF+Output+Plugin

WekaScoring

The Weka scoring plugin is a tool that allows classification and clustering models created with Weka to be used to "score" new data as part of a Kettle transform.

Source Location: https://github.com/pentaho/pdi-weka-scoring-plugin

Archtiectural Owner: Mark Hall

Architectural Area: Analytics / Reporting

Documentation Link: http://wiki.pentaho.com/display/DATAMINING/Using+the+Weka+Scoring+Plugin

Schema Workbench Plugins

pentaho-mondrianschemaworkbench-plugins

This plugin to Mondrian Schema Workbench allows for publishing of a Mondrian schema to the Pentaho Business Analytics Server.

Source Location: https://github.com/pentaho/pentaho-mondrianschemaworkbench-plugins

Architectural Owner: Will Gorman

Architectural Area: Analytics / Reporting