I am novice to Spring Cloud Data flow and Stream Cloud Streaming Applications.
Currently my project diagram looks like following :
I route a POST request from outside client using zuul API gateway to a microservice called Composite. Composite creates a stream using REST POST and deployes onto Spring Cloud Data Flow Server. As far as I know the microservices mongodb and file run as co-existing JVM processes. If My client has to know the status of stream, status of the processed data, How should Composite Microservice interact with Spring Cloud Data Flow Server? Currently when I make POST call to deploy the stream I dont even get the status from SCDF Server. Does SCDF expose any hooks to look at the individual apps? Also how can I change the flow #runtime to create a dynamic mesh?
Currently I am using Local Spring Cloud Data Flow Server for development.
Runtime platform is local
Local runtime is recommended only for development purpose and if you're preparing for production, please make sure to choose a platform variant (eg: cf, k8s, yarn, ..) that comes with non-functional requirements to support reliable and durable execution of all the applications running in streaming pipeline.
As far as I know the microservices mongodb and file run as co-existing JVM processes.
If your stream definition is file | mongodb, you'd have 2 different JVM's even when using Local runtime. They're independent Boot applications.
How should Composite Microservice interact with Spring Cloud Data Flow Server?
Not clear what you mean by "composite" here. All the microservice applications in SCDF communicate via messaging middleware such as Kafka or Rabbit. SCDF provides the orchestration capability to run such applications into various runtime platforms.
Currently when I make POST call to deploy the stream I dont even get the status from SCDF Server
You can use SCDF's REST-APIs to query for current status of the apps and it is platform agnostic. You can view the list of supported APIs by hitting the root URL (see image below) - there's a gap in docs - we will fix it. Following APIs could be useful for status checks.
Does SCDF expose any hooks to look at the individual apps?
Once the apps are deployed in a runtime platform, you can take advantage of Boot's actuator endpoints to explore more details such as trace, metrics, health, env among others at each application level. See Boot's actuator endpoints for more details. For instance, if your mongodb app is running locally and on port 23000, then you can check granular metrics for this application at: http://localhost:23000/metrics.
[As an FYI: future SCDF releases would include integrating Spring Boot + Spring Cloud Sleuth metrics and visual representation of the same.]
Also how can I change the flow #runtime to create a dynamic mesh?
If you're referring to editing a running streaming pipeline with addition/deletes, we are currently exploring design approach to support this functionality.
Related
I have an application which listen to an activeMQ queue and start a Batch Job when receiving a message.
I'd like to use Spring Cloud Dataflow to provide an UI but I can't find informations on how to configure it.
Since it uses Spring Boot I should be able to replicate how my application currently works (use a REST API to make it listen to activeMQ and start job when receiving message), but I can't find anything on how to make it start the batch in Cloud Dataflow.
You have a few options here.
Option 1: Launch your application as-is and manually send message to launch task.
Any arbitrary Spring Boot application can be launched from Dataflow (simply register it as type = "App").
Taken from https://github.com/spring-cloud/spring-cloud-dataflow/blob/main/spring-cloud-dataflow-docs/src/main/asciidoc/streams.adoc#register-a-stream-application:
Registering an application by using --type app is the same as registering a source, processor or sink. Applications of the type app can be used only in the Stream Application DSL (which uses double pipes || instead of single pipes | in the DSL) and instructs Data Flow not to configure the Spring Cloud Stream binding properties of the application. The application that is registered using --type app does not have to be a Spring Cloud Stream application. It can be any Spring Boot application. See the Stream Application DSL introduction for more about using this application type.
You would have to send the task launch in your code. You can use the Dataflow REST client to do this. You can get an idea of how to do that by looking at https://github.com/spring-cloud/spring-cloud-dataflow/tree/main/spring-cloud-dataflow-tasklauncher/spring-cloud-dataflow-tasklauncher-sink.
Option 2: Use pre-built stream applications to model the same flow as your application.
The app you describe can be logically modeled as a Spring Cloud Stream application.
There is a JMS source (provides messages to signal the need to kickoff task/batch job)
There is a TaskLauncher sink (receives messages and kicks off the task/batch job)
This app can actually be constructed w/ little effort by using the pre-packaged applications to model this flow.
JMS Source
Dataflow Tasklauncher Sink
If you have to register these applications in the UI - they can be found at:
maven://org.springframework.cloud.stream.app:jms-source-kafka:3.1.1
maven://org.springframework.cloud:spring-cloud-dataflow-tasklauncher-sink-kafka:2.9.2
Stream definition:
jms-source | dataflow-tasklauncher-sink
The README(s) on the above source/sinks give details about the configuration options.
Option 3: Custom Spring Cloud Stream app w/ function composition
The previous option would create 2 separate apps. However, if you want to keep the logic in a single app then you can look into creating a custom Spring Cloud Stream app that uses function composition and leverage the pre-built reusable Java functions that the apps in option 2 are built upon.
JMS Supplier
TaskLauncherFunction
I'm building an event-driven microservice architecture, which is supposed to be Cloud agnostic (as much as possible). Since this is initially going in GCP and I don't want to spend a long time in configurations and all that, I was going to use GCP's Pub/Sub directly for the event queue and would take care of other Cloud implementations later, but then I came across Spring Cloud Dataflow, which seemed nice because these are Spring Boot microservices and I needed a way to orchestrate them.
Does Spring Cloud Dataflow support Pub Sub as it's event queue?
Would it make my life easier in terms of configuration and setup going that path, rather than choosing a non native broker?
It'd be useful first to unpack the Spring Cloud Stream's "binder abstraction" because it is using this framework, you'd have a portable event-driven streaming application, which can run locally in your laptop or any cloud of your choice against the desired message broker.
Learn more about the binder-abstraction here. Here are all the available binder implementations of choice. Google PubSub is an option, and it is maintained by Google here.
Now, let's talk about Spring Cloud Data Flow (SCDF). Once when you have built the streaming applications, you could use SCDF to design+create a data pipeline made of such applications. There's the option to mix and reuse the collection of utility applications that we build, maintain, and release as well. The utility applications can be packaged with Google PubSub or other binders. More details here.
When you deploy the data pipeline, SCDF will resolve and download the individual applications to deploy them natively on platforms like Kubernetes or Cloud Foundry. We have users doing the same in a variety of cloud infrastructure (VMs, Bare-metal, EC2, Rackspace, etc.), including DIY platforms, too.
While also automating the deployment of the applications, SCDF will automate the configuration setup based on naming conventions derived from stream/task and application names as a combination. So, when the apps bootstrap, they would have automatically received the connection configurations (from SCDF) and as well the destination/topic to connect to along with the other metadata to reason through a collection of apps as a "stream" or a "task/batch" data pipeline. This allows you to monitor and manage the pipelines centrally.
Lastly, there's the native ability in SCDF to rolling-upgrade/rolling-downgrade 1 or many applications in a data pipeline without impacting the upstream or downstream consumers in production. More details here. There's a webinar recording (demo starts at ~41.25) on how to do with CI/CD automation.
I'm looking for the azure alternative for the Data flow model of Data Source-processor-sink.
I want the three entities to be separate microservices. I want to use messaging as a link between these three.
Basically, Source app takes the data from another service and sends it to processor while processor app acts on it and sends relevant notification/alert to sink.
I'm aware I can use rabbitmq for the messaging but I need to know which one will be better in azure - service bus topics or eventhub? and how can I use them?
At the moment, there isn't a Spring Cloud Stream binder implementation for Azure Event Hubs.
Unless we have this, the out-of-the-box or the custom apps cannot be built as a messaging-microservice app, where Spring Cloud Stream provides the programming model and Spring Cloud Data Flow lets you orchestrate the individual microserivces in to a data pipeline (i.e., source-processor-sink) via the DSL/Drag-and-Drop GUI.
Microsoft was exploring the binder implementation in the past; possibly it would end up in Azure Spring Boot project. Feel free to drop an issue on their backlog.
I am trying to work through a solution where the workflow is like this:
User hits a microservice to upload images
That microservice de-duplicates the image and if it really is new, queues it up for processing
The processing chain lives in Spring Cloud Dataflow
The microservice already exists, and we are trying to extend it to do the fancy processing. My initial cut was to use the Http Source from the sample starter pack since that would be something I didn't have to create. The problem is that the source doesn't register itself with Spring Discovery server, so there is no way to get an end point without making gross assumptions (like it lives on the dataflow server at port XYZ).
We can create a Queue endpoint and send the data directly a Queue source that receives the outside event and forwards it to an SCDF queue.
What would be awesome is if DataFlow could connect the start of the queue for me, without repackaging the microservice as a Source.
The major issue with Spring Data Flow is that it does not automatically start up deployed streams when the server starts up, and we need to be reasonably sure that microservice is always up.
The lifecycle of the server is decoupled from the apps it deploys, that was intentional.
I'm not following your thoughts on how dataflow could connect the start of the queue, but from your description there's a few things you could do:
You would need to modify the app in order to have it registered with eureka, but this is a very simple operation, no more than a few lines of code:
You can either start from a stream app perspective: https://start-scs.cfapps.io/ , select http source, your binder, and then add the spring-cloud-netflix library as well as #EnableDiscoveryClient at the Main boot class
Start with http://start.spring.io Select Stream Rabbit or Stream Kafka, add Web and netflix libraries, then add the #EnableDiscoveryClient and #EnableBinding annotations and create a simple HTTP endpoint for your use case.
In any case should be a small addition.
You can also open an issue at :https://github.com/spring-cloud-stream-app-starters/http/issues suggesting that we add #EnableDiscoveryClient to the http source app, we can take that in consideration on our next iteration as well.
I'll try to clarify few bits.
upload images -> if it really is new -> queues it up for processing
Upon a new upload event, you'd want to process the image. Here's a similar use-case, but more of a real-time streaming style solution. This is not what you're looking to do, but I thought it might be useful.
Porting the image processing code to a Spring Cloud Stream application is as simple as adding #EnableBinding(Processor.class). It is the same business logic - whether you're running it separately or orchestrating it via SCDF, it is still a standalone microservice. However, SCDF expects it to be either a Source, Processor, Sink, or Task application types. We will be opening this up to support any arbitrary "functions" (lambdas) in the future release.
We can create a Queue endpoint and send the data directly a Queue source that receives the outside event and forwards it to an SCDF queue.
This is one of the standard solutions. You can directly consume new events (images) from a queue/topic and process it in the image-processor that we created in previous step. The named-channel support in DSL facilitates just that.
What would be awesome is if DataFlow could connect the start of the queue for me, without repackaging the microservice as a Source.
I'm not sure I understand this. If I were to assume, you're looking for "named-channel" as source and that is supported.
The major issue with Spring Data Flow is that it does not automatically start up deployed streams when the server starts up, and we need to be reasonably sure that microservice is always up.
The moment you deploy a Stream in SCDF, all the individual steps included in the DSL (i.e., stream definition) are resolved and deployed as standalone apps in the target runtime (cloud foundry, kubernetes, etc.,). Once deployed, it is left to the platform where the apps run for lifecycle management. SCDF does not retain or track the app states.
We are planning to retire the existing legacy java batch applications and recreate it with the latest available batch framework.
Given that we have a large number of batch jobs to be modernised, we are looking for a framework or architecture that would allow us to
Develop a batch solution that would allow us to dynamically deploy a new batch as and when they are created, without disturbing the existing deployed applications. - Does Spring cloud Task provide any of this feature. Note: We are looking only to deploy the apps to our local server, and has nothing to do with cloud.
If Spring Batch/Boot can provide us the feature we typically expect from a batch application, what is the special value add to go for Spring Cloud Task? - I wasn't able to completely understand this from the Spring documentation available online.
From the documentation of the Spring Cloud Task, I was able to understand that it allows an application to have many tasks within it. What should I do if each of the tasks have their own library dependencies, which might contradict with the dependencies of other Tasks? So in that case, should each of these tasks moved to a new application or this there a work around for that?
To answer your questions:
Does Spring Cloud Task handle orchestration - No. Spring Cloud Task does not handle orchestration of tasks or jobs. The component in this ecosystem that handles the deployment/orchestration of tasks or jobs is really Spring Cloud Data Flow (which is why I asked if you use any type of cloud platform including YARN, Cloud Foundry, Kubernetes, or Mesos...the environments supported by Spring Cloud Data Flow).
What added value does Spring Cloud Task provide over Spring Boot/Spring Batch - Spring Cloud Task is designed to provide a few things:
Similar abilities to Spring Batch with regards to state management without needing to create a batch job. When running a Boot application on a cloud environment, there is no standard way of getting the results from environment to environment (YARN handles job results differently from tasks on Cloud Foundry which is different from jobs on Kubernetes, etc). Spring Batch provides this but now all short lived processes need the overhead of the Batch API so Spring Cloud Task provides a lighter touch to those use cases.
Automatically adds informational listeners. With Spring XD, when you ran a job in an XD container, the XD container automatically added a number of informational listeners that broadcast events that you could listen for. Spring Cloud Task brings the same functionality without the need for the XD container.
Integration with Spring Cloud Stream. Spring Cloud Task provides the ability to launch tasks from messages received from Spring Cloud Stream. Also, the informational messages previously mentioned (both Batch events as well as Task events) are sent via Spring Cloud Stream channels.
The DeployerPartitionHandler. When working in a cloud environment, this PartitionHandler implementation allows you to launch workers for a partitioned batch job as tasks. This allows for the dynamic scaling of partitioned batch jobs instead of the traditional option of pre-deploying workers that listen for work which wastes resources in a modern cloud environment.
How does the packaging of multiple tasks work with dependencies - In short, this is not recommended. The idea of a Spring Cloud Task is that the execution of the Spring Boot application is the Task. While you could package up multiple tasks and using different methods, have them execute based on different stimulus, that goes against the 12 factor application concepts which are essential for correct use of Spring Cloud Task.
My two cents
For the best option for a modern batch platform, you really need to look into some from of platform first and that begins at the Cloud Foundry/Kubernetes/Mesos/YARN layer. Without that, you end up building a large part of the infrastructure yourself. That is why Spring XD evolved into Spring Cloud Data Flow. The added complexity that lived in the containers of Spring XD is removed by requiring a modern platform to run on (since they all handle those guarantees themselves). Without that piece, you're going to spend a lot of time managing the deployment and orchestration of applications that most modern platforms handle for you.
From there, the choice becomes pretty easy IMHO with Spring Cloud Task for simple tasks, Spring Batch for batch jobs, and Spring Cloud Data Flow for orchestration.