This project provides support for orchestrating long-running (streaming) and short-lived (task/batch) data microservices to Marathon on Mesos.

Spring Cloud Data Flow is a cloud-native orchestration service for composable data microservices on modern runtimes. With Spring Cloud Data Flow, developers can create and orchestrate data pipelines for common use cases such as data ingest, real-time analytics, and data import/export.

The Spring Cloud Data Flow architecture consists of a server that deploys Streams and Tasks. Streams are defined using a DSL or visually through the browser based designer UI. Streams are based on the Spring Cloud Stream programming model while Tasks are based on the Spring Cloud Task programming model. The sections below describe more information about creating your own custom Streams and Tasks

For more details about the core architecture components and the supported features, please review Spring Cloud Data Flow’s core reference guide. There’re several samples available for reference.

Spring Cloud Stream is a framework for building message-driven microservice applications. Spring Cloud Stream builds upon Spring Boot to create standalone, production-grade Spring applications, and uses Spring Integration to provide connectivity to message brokers. It provides opinionated configuration of middleware from several vendors, introducing the concepts of persistent publish-subscribe semantics, consumer groups, and partitions.

For more details about the core framework components and the supported features, please review Spring Cloud Stream’s reference guide.

There’s a rich ecosystem of Spring Cloud Stream Application-Starters that can be used either as standalone data microservice applications or in Spring Cloud Data Flow. For convenience, we have generated RabbitMQ and Apache Kafka variants of these application-starters that are available for use from Maven Repo and Docker Hub as maven artifacts and docker images, respectively.

Do you have a requirement to develop custom applications? No problem. Refer to this guide to create custom stream applications. There’re several samples available for reference.

Spring Cloud Task makes it easy to create short-lived microservices. We provide capabilities that allow short-lived JVM processes to be executed on demand in a production environment.

For more details about the core framework components and the supported features, please review Spring Cloud Task’s reference guide.

There’s a rich ecosystem of Spring Cloud Task Application-Starters that can be used either as standalone data microservice applications or in Spring Cloud Data Flow. For convenience, the generated application-starters are available for use from Maven Repo. There are several samples available for reference.

In this getting started guide, the Data Flow Server is run as a standalone application outside the Mesos cluster. A future version will provide support for the Data Flow Server itself to run on Mesos.

In Spring Cloud Data Flow, a basic stream defines the ingestion of event driven data from a source to a sink that passes through any number of processors. Streams are composed of spring-cloud-stream applications and the deployment of stream definitions is done via the Data Flow Server (REST API). The Getting Started section shows you how to start these servers and how to start and use the Spring Cloud Data Flow shell.

A high level DSL is used to create stream definitions. The DSL to define a stream that has an http source and a file sink (with no processors) is shown below

http | file

The DSL mimics a UNIX pipes and filters syntax. Default values for ports and filenames are used in this example but can be overridden using -- options, such as

http --server.port=8091 | file --directory=/tmp/httpdata/

To create these stream definitions you use the shell or make an HTTP POST request to the Spring Cloud Data Flow Server. More details can be found in the sections below.

In the examples above, we connected a source to a sink using the pipe symbol |. You can also pass properties to the source and sink configurations. The property names will depend on the individual app implementations, but as an example, the http source app exposes a server.port setting which allows you to change the data ingestion port from the default value. To create the stream using port 8000, we would use

dataflow:> stream create --definition "http --server.port=8000 | log" --name myhttpstream

The shell provides tab completion for application properties and also the shell command app info provides some additional documentation.

Register a Stream App with the App Registry using the Spring Cloud Data Flow Shell app register command. You must provide a unique name, application type, and a URI that can be resolved to the app artifact. For the type, specify "source", "processor", or "sink". Here are a few examples:

dataflow:>app register --name mysource --type source --uri maven://com.example:mysource:0.0.1-SNAPSHOT

dataflow:>app register --name myprocessor --type processor --uri file:///Users/example/myprocessor-1.2.3.jar

dataflow:>app register --name mysink --type sink --uri http://example.com/mysink-2.0.1.jar

When providing a URI with the maven scheme, the format should conform to the following:

maven://<groupId>:<artifactId>[:<extension>[:<classifier>]]:<version>

For example, if you would like to register the snapshot versions of the http and log applications built with the RabbitMQ binder, you could do the following:

dataflow:>app register --name http --type source --uri maven://org.springframework.cloud.stream.app:http-source-rabbit:1.0.0.BUILD-SNAPSHOT
dataflow:>app register --name log --type sink --uri maven://org.springframework.cloud.stream.app:http-log-rabbit:1.0.0.BUILD-SNAPSHOT

If you would like to register multiple apps at one time, you can store them in a properties file where the keys are formatted as <type>.<name> and the values are the URIs.

For example, if you would like to register the snapshot versions of the http and log applications built with the RabbitMQ binder, you could have the following in a properties file [eg: stream-apps.properties]:

source.http=maven://org.springframework.cloud.stream.app:http-source-rabbit:1.0.0.BUILD-SNAPSHOT
sink.log=maven://org.springframework.cloud.stream.app:log-sink-rabbit:1.0.0.BUILD-SNAPSHOT

Then to import the apps in bulk, use the app import command and provide the location of the properties file via --uri:

dataflow:>app import --uri file:///<YOUR_FILE_LOCATION>/stream-apps.properties

For convenience, we have the static files with application-URIs (for both maven and docker) available for all the out-of-the-box Stream and Task app-starters. You can point to this file and import all the application-URIs in bulk. Otherwise, as explained in previous paragraphs, you can register them individually or have your own custom property file with only the required application-URIs in it. It is recommended, however, to have a "focused" list of desired application-URIs in a custom property file.

List of available static property files:

For example, if you would like to register all out-of-the-box stream applications built with the RabbitMQ binder in bulk, you can with the following command.

dataflow:>app import --uri http://bit.ly/stream-applications-rabbit-maven

You can also pass the --local option (which is TRUE by default) to indicate whether the properties file location should be resolved within the shell process itself. If the location should be resolved from the Data Flow Server process, specify --local false.

When using either app register or app import, if a stream app is already registered with the provided name and type, it will not be overridden by default. If you would like to override the pre-existing stream app, then include the --force option.

	Note
	In some cases the Resource is resolved on the server side, whereas in others the URI will be passed to a runtime container instance where it is resolved. Consult the specific documentation of each Data Flow Server for more detail.

configuration.classes=org.springframework.cloud.stream.app.file.sink.FileSinkProperties

configuration.classes=org.springframework.cloud.stream.app.file.sink.FileSinkProperties
configuration-properties.names=server.port

The Spring Cloud Data Flow Server exposes a full RESTful API for managing the lifecycle of stream definitions, but the easiest way to use is it is via the Spring Cloud Data Flow shell. Start the shell as described in the Getting Started section.

New streams are created by posting stream definitions. The definitions are built from a simple DSL. For example, let’s walk through what happens if we execute the following shell command:

dataflow:> stream create --definition "time | log" --name ticktock

This defines a stream named ticktock based off the DSL expression time | log. The DSL uses the "pipe" symbol |, to connect a source to a sink.

Then to deploy the stream execute the following shell command (or alternatively add the --deploy flag when creating the stream so that this step is not needed):

dataflow:> stream deploy --name ticktock

The Data Flow Server resolves time and log to maven coordinates and uses those to launch the time and log applications of the stream.

2016-06-01 09:41:21.728  INFO 79016 --- [nio-9393-exec-6] o.s.c.d.spi.local.LocalAppDeployer       : deploying app ticktock.log instance 0
   Logs will be in /var/folders/wn/8jxm_tbd1vj28c8vj37n900m0000gn/T/spring-cloud-dataflow-912434582726479179/ticktock-1464788481708/ticktock.log
2016-06-01 09:41:21.914  INFO 79016 --- [nio-9393-exec-6] o.s.c.d.spi.local.LocalAppDeployer       : deploying app ticktock.time instance 0
   Logs will be in /var/folders/wn/8jxm_tbd1vj28c8vj37n900m0000gn/T/spring-cloud-dataflow-912434582726479179/ticktock-1464788481910/ticktock.time

In this example, the time source simply sends the current time as a message each second, and the log sink outputs it using the logging framework. You can tail the stdout log (which has an "_<instance>" suffix). The log files are located within the directory displayed in the Data Flow Server’s log output, as shown above.

$ tail -f /var/folders/wn/8jxm_tbd1vj28c8vj37n900m0000gn/T/spring-cloud-dataflow-912434582726479179/ticktock-1464788481708/ticktock.log/stdout_0.log
2016-06-01 09:45:11.250  INFO 79194 --- [  kafka-binder-] log.sink    : 06/01/16 09:45:11
2016-06-01 09:45:12.250  INFO 79194 --- [  kafka-binder-] log.sink    : 06/01/16 09:45:12
2016-06-01 09:45:13.251  INFO 79194 --- [  kafka-binder-] log.sink    : 06/01/16 09:45:13

If you would like to have multiple instances of an application in the stream, you can include a property with the deploy command:

dataflow:> stream deploy --name ticktock --properties "app.time.count=3"

	Important
	See Chapter 16, Using Labels in a Stream.

You can delete a stream by issuing the stream destroy command from the shell:

dataflow:> stream destroy --name ticktock

If the stream was deployed, it will be undeployed before the stream definition is deleted.

Often you will want to stop a stream, but retain the name and definition for future use. In that case you can undeploy the stream by name and issue the deploy command at a later time to restart it.

dataflow:> stream undeploy --name ticktock
dataflow:> stream deploy --name ticktock

Let’s try something a bit more complicated and swap out the time source for something else. Another supported source type is http, which accepts data for ingestion over HTTP POSTs. Note that the http source accepts data on a different port from the Data Flow Server (default 8080). By default the port is randomly assigned.

To create a stream using an http source, but still using the same log sink, we would change the original command above to

dataflow:> stream create --definition "http | log" --name myhttpstream --deploy

which will produce the following output from the server

2016-06-01 09:47:58.920  INFO 79016 --- [io-9393-exec-10] o.s.c.d.spi.local.LocalAppDeployer       : deploying app myhttpstream.log instance 0
   Logs will be in /var/folders/wn/8jxm_tbd1vj28c8vj37n900m0000gn/T/spring-cloud-dataflow-912434582726479179/myhttpstream-1464788878747/myhttpstream.log
2016-06-01 09:48:06.396  INFO 79016 --- [io-9393-exec-10] o.s.c.d.spi.local.LocalAppDeployer       : deploying app myhttpstream.http instance 0
   Logs will be in /var/folders/wn/8jxm_tbd1vj28c8vj37n900m0000gn/T/spring-cloud-dataflow-912434582726479179/myhttpstream-1464788886383/myhttpstream.http

Note that we don’t see any other output this time until we actually post some data (using a shell command). In order to see the randomly assigned port on which the http source is listening, execute:

dataflow:> runtime apps

You should see that the corresponding http source has a url property containing the host and port information on which it is listening. You are now ready to post to that url, e.g.:

dataflow:> http post --target http://localhost:1234 --data "hello"
dataflow:> http post --target http://localhost:1234 --data "goodbye"

and the stream will then funnel the data from the http source to the output log implemented by the log sink

2016-06-01 09:50:22.121  INFO 79654 --- [  kafka-binder-] log.sink    : hello
2016-06-01 09:50:26.810  INFO 79654 --- [  kafka-binder-] log.sink    : goodbye

Of course, we could also change the sink implementation. You could pipe the output to a file (file), to hadoop (hdfs) or to any of the other sink apps which are available. You can also define your own apps.

As an example of a simple processing step, we can transform the payload of the HTTP posted data to upper case using the stream definitions

http | transform --expression=payload.toUpperCase() | log

To create this stream enter the following command in the shell

dataflow:> stream create --definition "http | transform --expression=payload.toUpperCase() | log" --name mystream --deploy

Posting some data (using a shell command)

dataflow:> http post --target http://localhost:1234 --data "hello"

Will result in an uppercased 'HELLO' in the log

2016-06-01 09:54:37.749  INFO 80083 --- [  kafka-binder-] log.sink    : HELLO

To demonstrate the data partitioning functionality, let’s deploy the following stream with Kafka as the binder.

dataflow:>stream create --name words --definition "http --server.port=9900 | splitter --expression=payload.split(' ') | log"
Created new stream 'words'

dataflow:>stream deploy words --properties "app.splitter.producer.partitionKeyExpression=payload,app.log.count=2"
Deployed stream 'words'

dataflow:>http post --target http://localhost:9900 --data "How much wood would a woodchuck chuck if a woodchuck could chuck wood"
> POST (text/plain;Charset=UTF-8) http://localhost:9900 How much wood would a woodchuck chuck if a woodchuck could chuck wood
> 202 ACCEPTED

You’ll see the following in the server logs.

2016-06-05 18:33:24.982  INFO 58039 --- [nio-9393-exec-9] o.s.c.d.spi.local.LocalAppDeployer       : deploying app words.log instance 0
   Logs will be in /var/folders/c3/ctx7_rns6x30tq7rb76wzqwr0000gp/T/spring-cloud-dataflow-694182453710731989/words-1465176804970/words.log
2016-06-05 18:33:24.988  INFO 58039 --- [nio-9393-exec-9] o.s.c.d.spi.local.LocalAppDeployer       : deploying app words.log instance 1
   Logs will be in /var/folders/c3/ctx7_rns6x30tq7rb76wzqwr0000gp/T/spring-cloud-dataflow-694182453710731989/words-1465176804970/words.log

Review the words.log instance 0 logs:

2016-06-05 18:35:47.047  INFO 58638 --- [  kafka-binder-] log.sink                                 : How
2016-06-05 18:35:47.066  INFO 58638 --- [  kafka-binder-] log.sink                                 : chuck
2016-06-05 18:35:47.066  INFO 58638 --- [  kafka-binder-] log.sink                                 : chuck

Review the words.log instance 1 logs:

2016-06-05 18:35:47.047  INFO 58639 --- [  kafka-binder-] log.sink                                 : much
2016-06-05 18:35:47.066  INFO 58639 --- [  kafka-binder-] log.sink                                 : wood
2016-06-05 18:35:47.066  INFO 58639 --- [  kafka-binder-] log.sink                                 : would
2016-06-05 18:35:47.066  INFO 58639 --- [  kafka-binder-] log.sink                                 : a
2016-06-05 18:35:47.066  INFO 58639 --- [  kafka-binder-] log.sink                                 : woodchuck
2016-06-05 18:35:47.067  INFO 58639 --- [  kafka-binder-] log.sink                                 : if
2016-06-05 18:35:47.067  INFO 58639 --- [  kafka-binder-] log.sink                                 : a
2016-06-05 18:35:47.067  INFO 58639 --- [  kafka-binder-] log.sink                                 : woodchuck
2016-06-05 18:35:47.067  INFO 58639 --- [  kafka-binder-] log.sink                                 : could
2016-06-05 18:35:47.067  INFO 58639 --- [  kafka-binder-] log.sink                                 : wood

This shows that payload splits that contain the same word are routed to the same application instance.

Taps can be created at various producer endpoints in a stream. For a stream like this:

stream create --definition "http | step1: transform --expression=payload.toUpperCase() | step2: transform --expression=payload+'!' | log" --name mainstream --deploy

taps can be created at the output of http, step1 and step2.

To create a stream that acts as a 'tap' on another stream requires to specify the source destination name for the tap stream. The syntax for source destination name is:

`:<stream-name>.<label/app-name>`

To create a tap at the output of http in the stream above, the source destination name is mainstream.http To create a tap at the output of the first transform app in the stream above, the source destination name is mainstream.step1

The tap stream DSL looks like this:

stream create --definition ":mainstream.http > counter" --name tap_at_http --deploy

stream create --definition ":mainstream.step1 > jdbc" --name tap_at_step1_transformer --deploy

Note the colon (:) prefix before the destination names. The colon allows the parser to recognize this as a destination name instead of an app name.

When a stream is comprised of multiple apps with the same name, they must be qualified with labels:

stream create --definition "http | firstLabel: transform --expression=payload.toUpperCase() | secondLabel: transform --expression=payload+'!' | log" --name myStreamWithLabels --deploy

One can connect to a specific destination name located in the broker (Rabbit, Kafka etc.,) either at the source or at the sink position.

The following stream has the destination name at the source position:

stream create --definition ":myDestination > log" --name ingest_from_broker --deploy

This stream receives messages from the destination myDestination located at the broker and connects it to the log app.

The following stream has the destination name at the sink position:

stream create --definition "http > :myDestination" --name ingest_to_broker --deploy

This stream sends the messages from the http app to the destination myDestination located at the broker.

From the above streams, notice that the http and log apps are interacting with each other via the broker (through the destination myDestination) rather than having a pipe directly between http and log within a single stream.

It is also possible to connect two different destinations (source and sink positions) at the broker in a stream.

stream create --definition ":destination1 > :destination2" --name bridge_destinations --deploy

In the above stream, both the destinations (destination1 and destination2) are located in the broker. The messages flow from the source destination to the sink destination via a bridge app that connects them.

If directed graphs are needed instead of the simple linear streams described above, two features are relevant.

First, named destinations may be used as a way to combine the output from multiple streams or for multiple consumers to share the output from a single stream. This can be done using the DSL syntax http > :mydestination or :mydestination > log.

Second, you may need to determine the output channel of a stream based on some information that is only known at runtime. In that case, a router may be used in the sink position of a stream definition. For more information, refer to the Router Sink starter’s README.

18.1 Common application properties

In addition to configuration via DSL, Spring Cloud Data Flow provides a mechanism for setting common properties to all the streaming applications that are launched by it. This can be done by adding properties prefixed with spring.cloud.dataflow.applicationProperties.stream when starting the server. When doing so, the server will pass all the properties, without the prefix, to the instances it launches.

For example, all the launched applications can be configured to use a specific Kafka broker by launching the configuration server with the following options:

--spring.cloud.dataflow.applicationProperties.stream.spring.cloud.stream.kafka.binder.brokers=192.168.1.100:9092
--spring.cloud.dataflow.applicationProperties.stream.spring.cloud.stream.kafka.binder.zkNodes=192.168.1.100:2181

This will cause the properties stream.spring.cloud.stream.kafka.binder.brokers and spring.cloud.stream.kafka.binder.zkNodes to be passed to all the launched applications.

	Note
	Properties configured using this mechanism have lower precedence than stream deployment properties. They will be overridden if a property with the same key is specified at stream deployment time (e.g. app.http.spring.cloud.stream.kafka.binder.brokers will override the common property).

Spring Cloud Data Flow provides a browser-based GUI which currently has 6 sections:

Upon starting Spring Cloud Data Flow, the Dashboard is available at:

http://<host>:<port>/dashboard

For example: http://localhost:9393/dashboard

If you have enabled https, then it will be located at https://localhost:9393/dashboard. If you have enabled security, a login form is available at http://localhost:9393/dashboard/#/login.

Note: The default Dashboard server port is 9393

Figure 19.1. The Spring Cloud Data Flow Dashboard

The Apps section of the Dashboard lists all the available applications and provides the control to register/unregister them (if applicable). By clicking on the magnifying glass, you will get a listing of available definition properties.

Figure 20.1. List of Available Applications

The Runtime section of the Dashboard application shows the Spring Cloud Data Flow cluster view with the list of all running applications. For each runtime app the state of the deployment and the number of deployed instances is shown. A list of the used deployment properties is available by clicking on the app id.

Figure 21.1. List of Running Applications

The Streams section of the Dashboard provides the Definitions tab that provides a listing of Stream definitions. There you have the option to deploy or undeploy those stream definitions. Additionally you can remove the definition by clicking on destroy.

Figure 22.1. List of Stream Definitions

The Create Stream section of the Dashboard includes the Spring Flo designer tab that provides the canvas application, offering a interactive graphical interface for creating data pipelines.

In this tab, you can:

Watch this screencast that highlights some of the "Flo for Spring Cloud Data Flow" capabilities. Spring Flo wiki includes more detailed content on core Flo capabilities.

Figure 23.1. Flo for Spring Cloud Data Flow

The Tasks section of the Dashboard currently has three tabs:

24.1 Apps

Apps encapsulate a unit of work into a reusable component. Within the Data Flow runtime environment Apps allow users to create definitions for Streams as well as Tasks. Consequently, the Apps tab within the Tasks section allows users to create Task definitions.

Note: You will also use this tab to create Batch Jobs.

Figure 24.1. List of Task Apps

On this screen you can perform the following actions:

• View details such as the task app options.
• Create a Task Definition from the respective App.

24.1.1 Create a Task Definition from a selected Task App

On this screen you can create a new Task Definition. As a minimum you must provide a name for the new definition. You will also have the option to specify various properties that are used during the deployment of the app.

Note: Each parameter is only included if the Include checkbox is selected.

24.1.2 View Task App Details

On this page you can view the details of a selected task app, including the list of available options (properties) for that app.

24.2 Definitions

This page lists the Data Flow Task definitions and provides actions to launch or destroy those tasks.

Figure 24.2. List of Task Definitions

24.2.1 Launching Tasks

Once the task definition is created, they can be launched through the Dashboard as well. Navigate to the Definitions tab. Select the Task you want to launch by pressing Launch.

On the following screen, you can define one or more Task parameters by entering:

• Parameter Key
• Parameter Value

Task parameters are not typed.

24.3 Executions

Figure 24.3. List of Task Executions

The Jobs section of the Dashboard allows you to inspect Batch Jobs. The main section of the screen provides a list of Job Executions. Batch Jobs are Tasks that were executing one or more Batch Job. As such each Job Execution has a back reference to the Task Execution Id (Task Id).

In case of a failed job, you can also restart the task. When dealing with long-running Batch Jobs, you can also request to stop it.

Figure 25.1. List of Job Executions

25.1 List job executions

This page lists the Batch Job Executions and provides the option to restart or stop a specific job execution, provided the operation is available. Furthermore, you have the option to view the Job execution details.

The list of Job Executions also shows the state of the underlying Job Definition. Thus, if the underlying definition has been deleted, deleted will be shown.

25.1.1 Job execution details

Figure 25.2. Job Execution Details

The Job Execution Details screen also contains a list of the executed steps. You can further drill into the Step Execution Details by clicking onto the magnifying glass.

25.1.2 Step execution details

On the top of the page, you will see progress indicator the respective step, with the option to refresh the indicator. Furthermore, a link is provided to view the step execution history.

The Step Execution details screen provides a complete list of all Step Execution Context key/value pairs.

	Important
	In case of exceptions, the Exit Description field will contain additional error information. Please be aware, though, that this field can only have a maximum of 2500 characters. Therefore, in case of long exception stacktraces, trimming of error messages may occur. In that case, please refer to the server log files for further details.

25.1.3 Step Execution Progress

On this screen, you can see a progress bar indicator in regards to the execution of the current step. Under the Step Execution History, you can also view various metrics associated with the selected step such as duration, read counts, write counts etc.

Figure 25.3. Step Execution History

The Analytics section of the Dashboard provided data visualization capabilities for the various analytics applications available in Spring Cloud Data Flow:

For example, if you have created the springtweets stream and the corresponding counter in the Counter chapter, you can now easily create the corresponding graph from within the Dashboard tab:

Using the icons to the right, you can add additional charts to the Dashboard, re-arange the order of created dashboards or remove data visualizations.