Easy Batch The simple, stupid batch processing framework for Java Mahmoud Ben Hassine https://benas.github.io @b_e_n_a_s

2 Agenda • Introduction • State of the art • Motivations • Easy Batch • Overview • Basic usage • Advanced topics • Wrap-up

3 Agenda • Introduction • State of the art • Motivations • Easy Batch • Overview • Basic usage • Advanced topics • Wrap-up

Batch vs Stream processing Batch processing Stream processing Bounded data set Unbounded data stream High latency Low latency Static data set Dynamic data stream

Batch processing • Long running jobs • No human interaction • No fancy GUIs • OutOfMemory errors! 5

State of the art* 6 JSR-352 Excellent solutions! But .. *: Big data tools like Spark, Flink, etc are out of scope

What’s wrong with
 Spring Batch / JSR 352? 7 “I have to admit I got a little overwhelmed by the complexity and amount of configuration needed for even a simple example”, Jeff Zapotoczny “What should we think of the Spring Batch solution? Complex. Obviously, it looks more complicated than the simple approaches. This is typical of a framework: the learning curve is steeper”, Arnaud Cogoluègnes “Recently evaluated Spring Batch, and quickly rejected it once I realized that it added nothing to my project aside from bloat and overhead”, RT. Person Complex configuration + Steep learning curve

What’s wrong with
 Spring Batch / JSR 352? 8 “The context of a Spring Batch application grows pretty quick and involves configuring a lot of stuff that, at the outset, it just doesn't seem like you should need to configure. A "job repository" to track the status and history of job executions, which itself requires a data source - just to get started? Wow, that's a bit heavy handed.”, Jeff Zapotoczny “On voit que l’on a besoin d’un transaction manager. Cette propriété est obligatoire, ce qui est à mon sens dommage pour les cas simples comme le nôtre où nous n’utilisons pas les transactions.”, Julien Jakubowski Mandatory components that you might not need “Spring Batch or How Not to Design an API.. Why do I Need a Transaction Manager? Why do I Need a Job Repository?”, William Shields

Agenda • Introduction • State of the art • Motivations • Easy Batch • Overview • Basic usage • Advanced topics • Wrap-up

Motivations (1/2) • Keep it simple, stupid • Flexible and extensible API • Modular architecture • Reduce boilerplate code 10 Build yet another: - big data - cloud-native - map-reduce - fault-tolerant - ultra high-performance - massively parallel - distributed - reactive - real-time - resilient - [put buzzword here]
 processing framework. No, this is not the goal.. Goals Non Goals

Motivations (2/2) #id,name,description,price,published,lastUpdate 0001,product1,description1,2500,true,2014-01-01 000x,product2,description2,2400,true,2014-01-01 0003,,description3,2300,true,2014-01-01 0004,product4,description4,-2200,true,2014-01-01 0005,product5,description5,2100,true,2024-01-01 0006,product6,description6,2000,true,2014-01-01,Blah! import java.util.Date; public class Product { private long id; private String name; private String description; private double price; private boolean published; private Date lastUpdate; // getters, setters omitted } products.csv Common requirements: - Read file line by line - Filter header record - Parse and map data to the Product bean - Validate product data - Do something with the product (business logic) - Log errors - Report statistics The goal is to keep focus on business logic! Boilerplate Product.java 11

Agenda • Introduction • State of the art • Motivations • Easy Batch • Overview • Basic usage • Advanced topics • Wrap-up

Easy Batch in a nutshell • Name: Easy Batch • Date of birth: 13/08/2012 • Weight: 108 Kb (v6) • DNA: https://github.com/j-easy/easy-batch 13

Overview 14

The Record abstraction (1/2) 15

public interface Record

{ /** * Header of the record. */ Header getHeader(); /** * Payload of the record. */ P getPayload(); } The Record abstraction (2/2) 16 Header (No, Source, etc) Payload (Raw Data) Record Multiple implementations: FlatFileRecord, XmlRecord, JsonRecord, JdbcRecord, JmsRecord, etc.. Record.java

The Batch abstraction 17 { record 1, record 2, ... record n } Batch public class Batch implements Iterable { private List records; } Batch.java

The Job abstraction 18 public interface Job extends Callable { String getName(); } class BatchJob implements Job { } • Synchronous execution JobReport report = jobExecutor.execute(job); • Asynchronous execution Future report = jobExecutor.submit(job); • Parallel execution jobExecutor
 .submitAll(job1, job2); • Scheduled execution scheduledExecutorService
 .schedule(job, 2, MINUTES);

Batch Jobs 19 • Read records in sequence • Process records in pipeline • Write records in batches

Workflow listeners Job listener 20

Workflow listeners Batch listener 21

Workflow listeners Reader/Writer listeners 22

Workflow listeners Pipeline listener 23

Reading data 24 • Streaming APIs • One record at a time • Hide low-level APIs

Filtering data 25 • Filter undesired records • Data cleaning • Filter chain: multiple filters

Mapping data 
 to domain objects 26 • POJO-centric development • Abstract data format • Enforce DDD

Validating data • Validate data against application’s constraints • Declarative approach: Bean Validation API (JSR303) public class Tweet { private int id; @NotNull private String user; @Size(min=0, max=280) private String message; } 27

Processing pipeline • Define application’s business logic • Multiple processors • Unix-like pipelines 28

Writing data 29 • Hide low-level APIs • Write records in batches • Transaction management for relational databases

Agenda • Introduction • State of the art • Motivations • Easy Batch • Overview • Basic usage • Advanced topics • Wrap-up

Demo 1

Demo 2

Agenda • Introduction • State of the art • Motivations • Easy Batch • Overview • Basic usage • Advanced topics • Wrap-up

Parallel processing 34 • Jobs are Callable objects => jobExecutor.submitAll(job1, job2) • ReportMerger API to merge partial reports • Suitable for physical/logical partitioning

Fault tolerance 35 • Retry feature • Retryable record reader/processor/writer • Custom RetryPolicy + RetryTemplate if needed • Skip feature • Batch scanning in case of write error • Skip bad records instead of failing the whole job

Real time monitoring 36 Job job = new JobBuilder().enableJmx(true).build();

Agenda • Introduction • State of the art • Motivations • Easy Batch • Overview • Basic usage • Advanced usage • Wrap-up

Wrap-up • Lightweight, free and open source • Easy to learn, configure and use • Flexible & extensible API • Modular architecture • Fault tolerance features • Declarative data validation • Real-time monitoring 38 • No step concept with flows • No remote partitioning • No remote chunking • Not suitable for big data The not so good ones The good ones

FAQs 39 • How does Easy Batch compare to Spring Batch? • Why does Easy Batch not persist job state in a database like Spring Batch? • Why does Easy Batch not provide a Step concept like Spring Batch?

Final word: be pragmatic! 40

Who is using Easy Batch? 41 “I use this framework in production (and love it)” chsFleury / @github “Try EasyBatch. The simple stupid Batch framework. Try it once and use it forever.” Eddy Bayonne / @stackoverflow “Loving it so far. Making something I'm working on very simple” zackehh_ / @twitter “Thanks @easy_batch. You guys rock - especially your use of fluent interfaces in your APIs :-) #cleancode” NorthConcepts / @twitter “we have successfully used @easy_batch in production at Leroy Merlin and we love it” benensi / @twitter Community feedback Trusted by

Thank you! https://benas.github.io @b_e_n_a_s https://github.com/j-easy/easy-batch