to microservices and beyond

© Copyright 2014 Pivotal. All rights reserved.© Copyright 2014 Pivotal. All rights reserved.

To Microservices and Beyond!

1

Simon Elisha (@simon_elisha) - CTO & Director of Field Engineering

© Copyright 2014 Pivotal. All rights reserved. 2

IT SEEMS THERE’S SOME HYPE…

M!CR0S3RV!C3$!!!!!

© Copyright 2014 Pivotal. All rights reserved.

DEFINE: Microservice

3

Loosely coupled service oriented architecture with bounded contexts

If every service has to be updated in concert, it’s not loosely coupled!

If you have to know about surrounding services you don’t have a bounded context.


Not Traditional (ESB-centric) SOA

Enterprise Service Bus

Service Service Service Service

Service Service Service Service

UI UI


But Microservices!


But why?


Issues We’ll Confront•Microservices are not an inherently superior architecture.

•We’re still building big systems from smaller things.

• Just like Docker won’t save the world, neither will microservices. They’re not free.

• You absolutely cannot forget about data.

• Let’s begin!


Not an end in themselves…It’s about Continuous Delivery!


What is Continuous Delivery?

9

$Busin

ess

Devel

opm

ent

QA

Ope

ratio

nsCu

stom

er


What is Continuous Delivery?

10

$


Keep the Wheel Spinning!

Design Develop

Test

Customer Feedback Customer

Delivery

Analytics


Prod Release #1

Prod Release #2

Prod Release #3

Agile Development

Waterfall Organization!


Silo Delivery

13

Project Mgmt UX Dev QA DBA Sys

AdminNet

AdminStorage Admin

Adapted from: http://www.slideshare.net/adriancockcroft/goto-berlin

MONOLITHIC DELIVERY


Continuous Delivery

14

Product Mgr UX Dev QA DBA Sys

AdminNet

AdminStorage Admin

BUSINESS CAPABILITY TEAMS USING MICROSERVICES

PLATFORM OPERATIONS TEAM



Continuous Delivery

15

Product Mgr UX Dev QA DBA Sys

AdminNet

AdminStorage Admin

BUSINESS CAPABILITY TEAMS USING MICROSERVICES

PLATFORM OPERATIONS TEAM


Self Service

API


INVENTORY

Prod Release

Prod Release

Prod Release

CATALOG

Prod Release

Prod Release

Prod Release

REVIEWS

Prod Release

Prod Release

Prod Release

SHIPPING

Prod Release

Prod Release

Prod Release


Microservices Enabling Continuous Delivery•Decoupling Capabilities -> Decoupling Change Cycles

• Product Ownership: Tip to Tail

•We Build and Operate What We Understand BEST

•We Collaborate via API Contracts

•Microservice to Microservice

•Microservice to Platform


Systems over ServicesComposition over Components


Microframeworks for Microservices

19

Spring Boot http://projects.spring.io/spring-boot

Dropwizard http://dropwizard.io


It can get pretty small…

20

@RestController class ThisWillActuallyRun { @RequestMapping("/") String home() { Hello World! } }


With Spring Data REST!

21

http://projects.spring.io/spring-data-rest

@Entity @Table(name = "city") public class City implements Serializable {

@Id @GeneratedValue private Long id;

@Column(nullable = false) private String name;

@Column(nullable = false) private String county;

//...

}



22


@RepositoryRestResource(collectionResourceRel = "cities", path = "cities") public interface CityRepository extends PagingAndSortingRepository<City, Long> {}

@Configuration @ComponentScan @EnableAutoConfiguration @EnableJpaRepositories @Import(RepositoryRestMvcConfiguration.class) public class Application {

public static void main(String[] args) { SpringApplication.run(Application.class, args); } }



23


{ "_links" : { "next" : { "href" : "http://localhost:8080/cities?page=1&size=20" }, "self" : { "href" : "http://localhost:8080/cities{?page,size,sort}", "templated" : true } }, "_embedded" : { "cities" : [ { "name" : "HOLTSVILLE", "county" : "SUFFOLK", "stateCode" : "NY", "postalCode" : "00501", "latitude" : "+40.922326", "longitude" : "-072.637078",


But No Microservice is an Island…

24


Example Distributed System: Minified

25


Some emergent features of microservices systems…

•Distributed/Versioned Configuration

• Service Registration/Discovery

• Routing/Load Balancing

• Fault Tolerance (Circuit Breakers, Bulkheads, …)


Example: Coordination Boiler Plate

27


• Eureka

•Hystrix + Turbine

• Ribbon

• Feign

• Zuul

•Archaius

+

http://netflix.github.io


http://projects.spring.io/spring-cloud


Example: Spring Cloud + Netflix OSS

30


Operationalized ArchitectureYou have to pay for your lunch!

© Copyright 2014 Pivotal. All rights reserved.© Copyright 2014 Pivotal. All rights reserved. 32

http://highscalability.com/blog/2014/4/8/microservices-not-a-free-lunch.html


Paying for your lunch…• Significant Operations Overhead

• Substantial DevOps Skills Required

• Implicit Interfaces

•Duplication of Effort

•Distributed System Complexity

•Asynchronicity is Difficult!

• Testability Challenges


You must be this tall to use Microservices…

34

http://martinfowler.com/bliki/MicroservicePrerequisites.html

https://www.flickr.com/photos/gusset/3723961589

• RAPID PROVISIONING

•BASIC MONITORING

• RAPID APPLICATION DEPLOYMENT

•DEVOPS CULTURE


It takes a platform…

35

http://techblog.netflix.com/2013/01/optimizing-netflix-api.htmlhttp://techblog.netflix.com/2014/03/the-netflix-dynamic-scripting-platform.html


It takes a platform…

36

Pivotal CF Spring Cloud


Platform Features• Environment Provisioning

•On-Demand/Automatic Scaling

• Failover/Resilience

• Routing/Load Balancing

•Data Service Operations

•Monitoring


Environment Provisioning/App Deployment

38

Rou

ter

Cloud Controller Service Broker Node(s)

DEA DEA

DEA DEA

Blobstore DB

Runtime

1. Upload bits/metadata

2. Create/bind services

3. Stage via Buildpack

4. Deploy via Container

push app+ app MD

SC

+ =

http://docs.cloudfoundry.org/devguide/deploy-apps/


Environment Provisioning/App Deployment

39

Rou

ter

Cloud Controller

DEA

Blobstore DB

Runtime

Detect Compile UploadNo

System Buildpacks

+ =

Yes

http://docs.cloudfoundry.org/buildpacks/


Deployment/Load Balancing/Scaling

40

Blobstore

R

oute

r Cloud Controller

DEA

Messaging (NATS)

DEA DEA

Runtime

Access App

http://docs.cloudfoundry.org/concepts/architecture/warden.html

Container Container


Failover/Resilience

41

Blobstore

R

oute

r Cloud Controller

DEA

Messaging (NATS)

DEA DEA

Runtime

Health ManagerActual StateDesired State

http://docs.cloudfoundry.org/concepts/architecture/#hm9k

Container Container Container


Data Service Operations: Cluster Provisioning

42

Blobstore

BOSH

Health Monitor

DBDeploy my Services

IaaS

Worker VMs BOSH Director

NATS Cassandra Node

Target VMCassandra Node

Target VMCassandra Node

Target VM

http://bosh.cloudfoundry.org/


DB

Rou

ter

Data Service Operations: Application Binding

43

Service credentials

reserve resources

obtain connection data

CLI Cloud Controller

Service Broker

Data Service

Runtime

create service (HTTP)

bind service (HTTP)

create service (HTTP)

bind service (HTTP)

http://docs.cloudfoundry.org/services/api.html


Monitoring

44


It’s All About the DataWhat about the BIG QUESTIONS?


This won’t work…

46


Instead!


Bounded Contexts

48

Movie Movie

Actor

Genre

Media Type

Media Type

Kiosk

Location

Media

Product Catalog

Inventory


Polyglot Persistence

49

REST XYou shall not pass…


But I have a question!

50

REST XYou shall not pass…

?

?

? ?

?

?


Lambda Architecture

51


Join via Events!

52


http://projects.spring.io/spring-xd/


Redbox Conceptual Workflow

55


SpringBox Microservices

56

Catalog Service

Inventory Service

Kiosk

Kiosk

Kiosk

Reservation Service

https://github.com/cf-platform-eng/springbox-datacloud


…and if you have a question:

57

Kiosk

Kiosk

Kiosk

Speed Layer

Batch Layer Serving Layer

Event Ingest

https://github.com/cf-platform-eng/springbox-datacloud

“What movie genres are most popular in what geographic locations?”



58

// Aggregate preferences across all kiosks: stream create --name kiosk_agg_prefs --definition "rabbit --queues=lambda.kiosk.events | field-value-counter --fieldName=genreIds"

// Tap aggregate preferences, filter for kiosk #1: stream create --name kiosk_1_prefs --definition "tap:stream:kiosk_agg_prefs > filter --expression=#jsonPath(payload,'$.locationId').equals(1) | field-value-counter --fieldName=genreIds"



59

field-value-counter display --name kiosk_agg_prefs field-value-counter display=kiosk_agg_prefs ------------------------------------------- - ----- VALUE - COUNT Action | 14 Adventure | 6 Comedy | 27 Sci-Fi | 18


Where We’ve Been…•Microservices are an enabler to Continuous Delivery.

• Less about services, more about composed distributed systems. Patterns can help.

• You’re going to need a platform.

•Decomposed data governance -> recomposed data discovery.

• Thank You!

A NEW PLATFORM FOR A NEW ERA

to microservices and beyond

Software

service oriented architecture

self service api

silo delivery

monolithic delivery

microservice microservice

api contracts microservice

simon elisha

bounded contexts