Pravega and Analytics Connectors Examples

This repository contains code samples to demonstrate how developers can work with Pravega. We also provide code samples to connect analytics engines such as Flink and Hadoop with Pravega as a storage substrate for data streams.

For more information on Pravega, we recommend to read the documentation and the developer guide.

Repository Structure

This repository is divided into sub-projects (pravega-client-examples, flink-connector-examples and hadoop-connector-examples), each one addressed to demonstrate a specific component. In these sub-projects, we provide a battery of simple code examples aimed at illustrating how a particular feature or API works. Moreover, we also include a scenarios folder that contains more complex applications as sub-projects, which show use-cases exploiting one or multiple components.

Hint: Have a look to the terminology and concepts in Pravega.

Pravega Client Examples

Example Name	Description	Language
gettingstarted	Simple example of how to read/write from/to a Pravega Stream.	Java
consolerw	Application that allows users to work with Stream, Transaction and StreamCut APIs via CLI.	Java
noop	Example of how to add a simple callback executed upon a read event.	Java
statesynchronizer	Application that allows users to work with StateSynchronizer API via CLI.	Java
streamcuts	Application examples demonstrating the use of StreamCuts via CLI.	Java

The related documentation and instructions are here.

Flink Connector Examples

Example Name	Description	Language
wordcount	Counting the words continuously from a Pravega Stream to demonstrate the usage of Flink connector for Pravega.	Java
primer	This sample demonstrates Pravega "exactly-once" feature jointly with Flink checkpointing and exactly-once mode.	Java
streamcuts	This sample demonstrates the use of Pravega StreamCuts in Flink applications.	Java

The related documentation and instructions are here.

Hadoop Connector Examples

Example Name	Description	Language
wordcount	Counts the words from a Pravega Stream filled with random text to demonstrate the usage of Hadoop connector for Pravega.	Java
terasort	Sort events from an input Pravega Stream and then write sorted events to one or more streams.	Java

The related documentation and instructions are here.

Scenarios

Example Name	Description	Language
turbineheatsensor	It emulates parallel sensors producing temperature values (writers) and parallel consumers performing real-time statistics (readers) via Pravega client.	Java
turbineheatprocessor	A Flink streaming application for processing temperature data from a Pravega stream produced by the turbineheatsensor app. The application computes a daily summary of the temperature range observed on that day by each sensor.	Java, Scala
anomaly-detection	A Flink streaming application for detecting anomalous input patterns using a finite-state machine.	Java
pravega-flink-connector-sql-samples	Flink connector table api/sql samples.	Java

Build Instructions

Next, we provide instructions for building the pravega-samples repository. There are two main options:

• Out-of-the-box: If you want a quick start, run the samples by building pravega-samples out-of-the-box (go straight to section Pravega Samples Build Instructions).
• Build from source: If you want to have fun building the different projects from source, please read section Building Pravega Components from Source (Optional) before building pravega-samples.

Pre-requisites

• Java 8

Building Pravega Components from Source (Optional)

Pravega Build Instructions

If you want to build Pravega from source, you may need to generate the latest Pravega jar files and install them to your local Maven repository. To build Pravega from sources and use it here, please run the following commands:

$ git clone https://github.com/pravega/pravega.git
$ cd pravega
$ ./gradlew install

The above command should generate the required jar files into your local Maven repository.

Hint: For using in the sample applications the Pravega version you just built, you need to update the pravegaVersion=<local_maven_pravega_version> property in gradle.properties file of pravega-samples.

For more information, please visit Pravega.

Flink Connector Build Instructions

To build the Flink connector from source, follow the below steps to build and publish artifacts from source to local Maven repository:

$ git clone --recursive https://github.com/pravega/flink-connectors.git
$ cd flink-connectors
$ ./gradlew install

Hint: For using in the sample applications the Flink connector version you just built, you need to update the flinkConnectorVersion=<local_maven_flink_connector_version> property in gradle.properties file of pravega-samples.

For more information, please visit Flink Connectors.

Hadoop Connector Build Instructions

To build the Hadoop connector from source, follow the below steps to build and publish artifacts from source to local Maven repository:

$ git clone --recurse-submodules https://github.com/pravega/hadoop-connectors.git
$ cd hadoop-connectors
$ ./gradlew install

Hint: For using in the sample applications the Hadoop connector version you just built, you need to update the hadoopConnectorVersion=<local_maven_hadoop_connector_version> property in gradle.properties file of pravega-samples.

For more information, please visit Hadoop Connectors.

Configuring Pravega Samples for Running with Source Builds

In the previous instructions, we noted that you will need to change the gradle.properties file in pravega-samples for using the Pravega components built from source. Here we provide an example of how to do so:

1. Imagine that we want to build Pravega from source. Let us assume that we executed git clone https://github.com/pravega/pravega.git and the last commit of master branch is 2990193xxx.

2. After executing ./gradlew install, we will see in our local Maven repository (e.g., ~/.m2/repository/io/pravega/*) artifacts that contain in their names that commit version such as 0.3.0-1889.2990193-SNAPSHOT. These artifacts are the result from building Pravega from source.

3. The only thing you have to do is to set pravegaVersion=0.3.0-1889.2990193-SNAPSHOT in the gradle.properties file of pravega-samples.

While this example is for Pravega, the same procedure applies for Flink and Hadoop connectors.

Pravega Samples Build Instructions

The pravega-samples project is prepared for working out-of-the-box with release artifacts of Pravega components, which are already available in Maven central. To build pravega-samples from source, use the built-in gradle wrapper as follows:

$ git clone https://github.com/pravega/pravega-samples.git
$ cd pravega-samples
$ ./gradlew clean installDist

That's it! You are good to go and execute the examples :)

To ease their execution, most examples can be run either using the gradle wrapper (gradlew) or scripts. The above gradle command automatically creates the execution scripts that can be found under:

pravega-samples/pravega-client-examples/build/install/pravega-client-examples/bin

There is a Linux/Mac script and a Windows (.bat) script for each separate executable.

Working with dev branch: If you are curious about the most recent sample applications, you may like to try the dev version of pravega-samples as well. To do so, just clone the dev branch instead of master (default):

$ git clone -b dev https://github.com/pravega/pravega-samples.git
$ cd pravega-samples
$ ./gradlew clean installDist

The dev branch works with Pravega snapshots artifacts published in our JFrog repository instead of using release versions.

Proposed Roadmap

We propose a roadmap to proceed with the execution of examples based on their complexity:

1. Pravega client examples: First step to understand the basics of Pravega and exercise the concepts presented in the documentation.
2. Flink connector examples: These examples show the basic functionality of the Flink connector for Pravega.
3. Hadoop connector examples: These examples show the basic functionality of the Hadoop connector for Pravega.
4. Scenarios: Applications that go beyond the basic usage of Pravega APIs, which may include complex interactions between Pravega and analytics engines (e.g., Flink, Hadoop, Spark) to demonstrate analytics use cases.

Where to Find Help

Documentation on Pravega and Analytics Connectors:

• Pravega.io, Pravega Wiki.
• Flink Connectors Wiki.

Did you find a problem or bug?

• First, check our FAQ.
• If the FAQ does not help you, create a new GitHub issue.

Do you want to contribute a new example application?

• Follow the guidelines for contributors.

Have fun!!