Azure Microservices in Practice - Radu Vunvulea

@ITCAMPRO #ITCAMP16Community Conference for IT Professionals

Azure Microservices in Practice

Radu Vunvulea

Solution Architect, iQuest Technologies

Microsoft Azure MVP

@RaduVunvulea



Many thanks to our sponsors & partners!

GOLD

SILVER

PARTNERS

PLATINUM

POWERED BY


RADU VUNVULEA MCTS MCP BANK HOME AUTOMATION

MVP ENTERPRISE

AUTOMOTIVE PHARMA

LEAN AND AGILE E-COMMERCE

WEB iQuest

AZURE JAVASCRIPT VUNVULEARADU.BLOGSPOT.COM

MOBILE DOTNET @RaduVunvulea

WCF WPF ENTHUSIASTIC


Scope


WHAT ARE MICROSERVICES?



Software architecture style in which complex

applications are composed of small, independent

processes communicating with each other using

language-agnostic APIs

Microservices


WHY WE SHOULD USE

MICROSERVICES?


If we need a system that has components that are…

High decoupled

Small

Easy to replace

Focus on doing a small task

Independent deployment units

Symmetrical architecture


Evolution

Maintenance

is hard

Adding new

features is slow

Changes are

expensive

Technology

becomes the

obstacle rather

than enabler

App Modernization


(AZURE) SERVICE FABRIC


• Service Fabric

• Azure Service Fabric

Why we have two different things?

?


Service Fabric | Azure Service Fabric

Microservices

Cloud

AzureAzure Service

Fabric

AWS Service Fabric

On-premises Service Fabric


Overview

Service Fabric

Azure

Windows Server

Linux

On-premises

Windows Server

Linux

Hosted clouds

Windows Server

Linux

Microservice

Microservice

Microservice

Microservice

Microservice


SERVICE TYPE


Rel

iab

le S

ervi

ce • Classical service

• Reliable

• Available

• Scalable

Act

or • State

• Turn-base access

• Single thread

• Reentrancy


Stat

eles

s •No state persisted between calls

Stat

efu

l

•‘A’ state is persisted between calls



What can we run in Service Fabric?

Service Fabric

Web Sites

Application

Custom logic


• Multiple programing languages (C# & Java)

• Support different commucation mechanisms:

–HTTP(s)

–WCF

–TCP

–UDP

• Any protocol and communication stack is supported

• Capable to run a guest executable

How



SERVICE LIFECYCLE


Reliable

Services Actors

Stateful

Stateless

State

StateState

First call…

Reliable Collections


Reliable

Services Actors

Stateful

Stateless

State

State

… after the call is executed

State


Reliable

Services Actors

Stateful

Stateless

State

State

… when GC is triggered

State



One UI for all environments


DEBUGGING AND TRACING


• Local debugging like any other .NET application

• Attach debugger to any Service Fabric Application

(on-premises or cloud)

• Diagnostic Events persistent in Cluster

• Real time Streaming Traces (Diagnostics Events)

Debugging and tracing



• Low latency

• High-throughput

• Transactional

• Can be used only

with Reliable Services

Reliable Collections

Dictionary

and

Queue

Replicated

Persisted

In-memory <<servic

e instance

>>

Asynchronous

Transactional


Where collection are stored


• Different instances of the same

service can access the same

Reliable Collection

• Instances from different

services cannot access the

same collection

Reliable Collection – visibility



TIMERS AND REMINDERS


Timers and Reminders

Timers

Similar with .NET

Called at a specific time interval

Reminders

Timer with persistence

Trigered in all circumstances (deactivations and failovers)

Concurrency

Turn based concurrency

Stateful Actos only


• Context: Multiple calls to the same Reliable Actor

Concurrency


Concurrency


Concurrency


Execution flow:

• Reminder Trigger

• SetPrice

• SetStock

Concurrency



Events

• Send events and

notifications from

services to caller

• Allows us be notified at

different steps on when

an exception occurs

• Can be used only for

actor-client

communication


• On Actor:

– IActorEvents

– Trigger our custom events

• On source:

– Implement callback (Event Handler)

– Subscribe our Event Handler

Events



TRACING AND DEBUGING


BEHIND THE SCENE


Partition and Replication


Fault Domains and Upgrade Domains


Where we can control them

Application

• Partitioning

• Replication

Cluster

• Fault Domain

• Upgrade Domain

• Naming Service

• Resource Balancer

• Failover Manager



Question

Answers


{“name” : “Radu Vunvulea,“blog” : “vunvulearadu.blogspot.com”,

“email” : ”[email protected]”,“socialMedia” :

{“twitter” : “@RaduVunvulea”,“fb” : “radu.vunvulea”

}

}


Today: Enjoy ITCamp Day 1Tomorow: Enjoy ITCamp Day 2

Saturday: Take a look on https://github.com/Azure/servicefabric-samples

https://github.com/Azure/servicefabric-samples

https://github.com/Azure/servicefabric-samples


{“name” : “Radu Vunvulea,“blog” : “vunvulearadu.blogspot.com”,

“email” : ”[email protected]”,“socialMedia” :

{“twitter” : “@RaduVunvulea”,“fb” : “radu.vunvulea”

}

}