Choreography vs Orchestration in serverless microservices Mete Atamel Developer Advocate at Google @meteatamel atamel.dev speakerdeck.com/meteatamel

Choreography vs Orchestration

Imagine a simple e-commerce transaction Services calling each other directly Frontend App Engine Order request Payment Processor Cloud Run Authorize & charge CC Shipper Cloud Functions Prepare & ship items Notifier Cloud Run Notify user

Simple REST: Pros and Cons Pros ➕ Better than a single monolith ➕ Easy to implement: Services simply call each other Cons ➖ Too much coupling ➖ Each service can be a single point of failure ➖ Each service needs its own error / retry / timeout logic ➖ Who ensures the whole transaction is successful? (hint: saga pattern)

Choreography (event-driven) Event-driven services Frontend App Engine Order request Payment Processor Cloud Run Authorize & charge CC Shipper Cloud Functions Prepare & ship items Notifier Cloud Run Notify user Message Broker

Choreography: Pros and Cons Pros ➕ Services are loosely coupled ➕ Services can be changed/scaled independently ➕ No single point of failure ➕ Events are useful to extend the system Cons ➖ Difficult to monitor ➖ Errors / retries / timeouts are hard ➖ The business flow is not captured explicitly ➖ Who ensures the whole transaction is successful?

Imagine a more complex transaction

Orchestration Orchestrated services Frontend App Engine Order request Payment Processor Cloud Run Authorize & charge CC Shipper Cloud Functions Prepare & ship items Notifier Cloud Run Notify user Orchestrator

Orchestration: Pros and Cons Pros ➕ Business flow captured centrally and source controlled ➕ Each step can be monitored ➕ Errors / retries / timeouts are centralized ➕ Services are still independent Cons ➖ A new orchestrator service to learn and maintain ➖ Orchestrator could be a single point of failure ➖ Loss of eventing flexibility ➖ How do you compensate for failed steps? (part of saga pattern)

Choreography or Orchestration?

It depends... Choreography Services are not closely related or can exist in different bounded contexts Orchestration Are services closely related? Can you describe the business logic in a flow chart?

Hybrid approach Orchestrated bounded contexts communicating via events Orchestrated Bounded Context Message Broker Orchestrated Bounded Context Orchestrated Bounded Context

Landscape

Choreography (event-driven) AWS: SQS, SNS, EventBridge Azure: Event Grid, Event Hubs, Service Bus Google Cloud: Pub/Sub, Eventarc Other: Kafka, Pulsar, Solace PubSub+, RabbitMQ, NATS...

Orchestration AWS: Step Functions Azure: Logic Apps Google Cloud: Workflows, Cloud Composer Other: Apache Airflow

AWS Step Functions Azure Logic Apps Workflows Organization AWS Microsoft Google Data serialization JSON JSON JSON and YAML Workflow metadata structure Yes? Yes? No? Named steps Yes Yes Yes Basic jumps Yes Yes Yes Step execution states Yes? Yes? No? Workflow execution state Yes Yes Yes Conditional jumps Yes Yes Yes HTTP Requests No Yes Yes Runtime parameters Yes Yes Yes Static variables definitions Yes Yes Yes Pass variable between steps Yes Yes Yes Data types supported JSON type String,Int,Double,Bool,Array,Ob ject,SecureString,SecureObject String, Int, Double, Bool, Array, Map/Object,Null Data type definition Implicit (JSON) Implicit Implicit JSON -> Dictionary Yes Yes Yes Arrays Yes Yes Yes Dictionaries Yes Yes Yes Array iterations Yes Yes Yes Parallel Array iteration Yes Yes No Built-in for(i =0;i<10;i++) Yes Yes Preview Parallel step execution Yes Yes No

Subworkflows No Yes Yes Expressions - math operators Yes Yes Yes Expressions - string functions Yes Yes Yes Expressions - other functions (SHA..) No Yes No Embedded JS No Yes No Environment Variables No Yes Limited Callbacks Yes Yes Preview Secrets No Yes Yes Bash commands No No No Delay Yes Yes Yes OnError Yes Yes Yes Retry Yes Yes Yes Return output Yes Yes Yes Input / output from a local file No No No echo to stdout No No No Build and run docker container steps No No No Open source runtime No No No User Interface - visualization Yes Yes Yes User Interface - visual editing Yes Yes No Connectors library Some Yes Yes AWS Step Functions Azure Logic Apps Workflows

CNCF Serverless Workflow serverlessworkflow.io Defines a vendor-neutral, open-source, and fully community-driven ecosystem for defining and running DSL-based workflows that target the Serverless technology domain. ● Specification for defining DSL-based workflows ● Developer SDKs for different programming languages ● Workflow runtimes supporting the specification ● Developer tooling support for writing DSL-based workflows

Orchestration: Google Cloud Workflows

Serverless Compute External API’s Google API’s etc... Workflows - orchestrate & integrate SaaS API’s Private API’s Other Clouds github.com/GoogleCloudPlatform/workflows-demos/tree/master/service-chaining

- processPayment: params: [paymentDetails] call: http.post args: url: https://payment-processor.run.app/... body: input: ${paymentDetails} result: processResult - shipItems: call: http.post args: url: https://.../cloudfunctions.net/ship body: input: ${processResult.body} result: shipResult - notifyUser: call: http.post ... Payment Processor Cloud Run Authorize & charge CC Notifier Cloud Run Notify user Shipper Cloud Functions Prepare & ship items YAML or JSON syntax

Payment Processor Cloud Run Authorize & charge CC Notifier Cloud Run Notify user Shipper Cloud Functions Prepare & ship items Payment Processor Cloud Run Authorize & charge CC Notifier Cloud Run Notify user Shipper Cloud Functions Prepare & ship items WAIT Payment Processor Cloud Run Authorize & charge CC Notifier Cloud Run Notify user Shipper Cloud Functions Prepare & ship items shipmentDetails userDetails Step Sequencing Serverless Pause Variable passing JSON Parsing Steps

Errors and retries Payment Processor Cloud Run Authorize & charge CC Notifier Cloud Run Notify user Shipper Cloud Functions Prepare & ship items MAX: 5 times BACKOFF Payment Processor Cloud Run Authorize & charge CC Notifier Cloud Run Notify user Shipper Cloud Functions Prepare & ship items Pager Cloud Run Escalate to support SUCCESS ERROR Configurable retries Configurable exception handling

Conditionals and 3rd party calls Notifier Cloud Run Notify user Shipper Cloud Functions Prepare & ship items Pager Cloud Run Escalate to support SUCCESS ERROR Out of Stock? No Request from the supplier Yes Read inventory Inventory DB Update inventory Inventory DB Supplier API

Other useful features Subworkflows to encapsulate common reusable flows Connectors to connect to other Google Cloud services & APIs More iterations, callbacks

Deploy, execute, manage workflows # Deploy a workflow gcloud workflows deploy my-workflow --source=workflow.yaml # Execute a workflow gcloud workflows execute my-workflow # See the result gcloud workflows executions describe --workflow my-workflow

No content

Case study: Pic-a-daily, A microservice-based picture sharing application

Pic-a-daily: A photo sharing application g.co/codelabs/serverless-workshop

Choreographed (event-driven) architecture

3 different event formats

Orchestrated architecture

Lessons Learned ➕ Simple REST was refreshing (vs. 3 eventing formats) ➕ Less code (eg. no event parsing, no Image Analysis & Garbage Collector functions) ➕ Less setup (eg. no Pub/Sub, no Scheduler, no Eventarc) ➕ Easier error handling (eg. the whole chain stops on error)

Lessons Learned ➖ New service to learn with its quirks and limitations ➖ Code vs. YAML, a single YAML file ➖ Debugging / testing / logging, no IDE support ➖ Loss of parallelism & eventing flexibility

Orchestration patterns

Choreography + Orchestration github.com/GoogleCloudPlatform/eventarc-samples/tree/main/processing-pipelines/image-v3

Retries and saga pattern github.com/GoogleCloudPlatform/workflows-demos/tree/master/retries-and-saga

Parallel steps and iteration github.com/GoogleCloudPlatform/workflows-demos/tree/master/twitter-sentiment-parallel Parallelize steps or iterations when you can!

Serverless + Serverful You can combine Serverless orchestration and VMs for best of both worlds 1. Containerize the long-running task, so it can run anywhere. 2. Plan to run the container on a Compute Engine VM with no time limitations. 3. Automate the creation of the VM, running of the container on the VM, and deletion of the VM with Workflows. github.com/GoogleCloudPlatform/workflows-demos/tree/master/long-running-container

Cloud Workflows cloud.google.com/workflows Pic-a-Daily Serverless Workshop g.co/codelabs/serverless-workshop Workflows samples github.com/GoogleCloudPlatform/workflows-demos Thank you! Mete Atamel @meteatamel atamel.dev speakerdeck.com/meteatamel