Ninja Azure Consultant Microservices? Anton Boyko Microsoft Azure MVP Ninja Azure Consultant boyko.ant@live.com
Agenda What and why? Several examples Implementation checklist What is microservice architecture approach and why should I care? Several examples Several examples of microservice architecture from real life projects Implementation checklist Several important points you need to consider while planning your microservice architecture What about Azure? What Microsoft Azure can offer for my microservices?
What and why?
Criteria Maintainability Monitoring Scalability Updates Onboarding How easy is it to maintain the code? How easy is it to fix a bug? Monitoring How easy is it to monitor solution health? Scalability How easy is it to add new computin power and handle heavier load? Updates How easy is it to update solution to the newest version? Onboarding How easy is it to get onboard a new team member?
Spaghetti $conn = new mysqli($servername, $username, $password, $dbname); if ($conn->connect_error) { die("Connection failed: " . $conn->connect_error); } $sql = "SELECT firstname, lastname FROM users"; $result = $conn->query($sql); if ($result->num_rows > 0) { echo "<table>"; while($row = $result->fetch_assoc()) { echo "<tr><td>"; echo $row["firstname"] . "</td><td>" . $row["lastname"]; echo "</td></tr>"; echo "</table>"; } else { echo "0 results"; $conn->close();
Spaghetti Maintainability Monitoring Scalability Updates Onboarding Impossible Monitoring Almost impossible Scalability Sometimes possible Updates Requires shutdown Onboarding Highly demotivating
Monolith Application in which user interface and data access code are combined into a single program. Design philosophy is that the application is responsible not just for a particular task, but can perform every step needed to complete a particular function.
Monolith Solution Presentation Application Data NLB Server Server
Monolith Maintainability Monitoring Scalability Updates Onboarding Mostly possible Highly depends on solution size Monitoring Possible Scalability Updates Requires shutdown Onboarding
Microservices Microservices is a variant of the service-oriented architecture (SOA) architectural style that structures an application as a collection of loosely coupled services. Microservices should be fine-grained and protocols should be lightweight. Application is easier to understand, develop and test.
Microservices Microservice 1 Presentation Application Data NLB Server Server Server Presentation
Microservices Maintainability Monitoring Scalability Updates Possible Monitoring Highly depends on solution size Scalability Updates Can be handled without shutdown Onboarding
Nanoservices Nanoservices is a microservices anti-pattern where each service is too fine grained. Nanoservice is a service whose overhead (communications, maintenance etc.) outweighs its utility.
Nanoservices User login Get categories Get goods Get my orders Presentation Presentation Presentation Presentation Application Application Application Application Data Data Data Data User register Add category Get good by id Create order Presentation Presentation Presentation Presentation Application Application Application Application Data Data Data Data
Nanoservices Maintainability Monitoring Scalability Updates Onboarding Almost impossible Monitoring Scalability Possible Updates Can be handled without shutdown Extremely depends on solution size Onboarding Sometimes possible
microservices = architecture
microservices ≠ golden hammer
SEVERAL examples
Online shop Identity Client Profile Admin Search Data Ad Social
Community portal Web Admin Rendering Identity Data Member Notification Speaker
Implementation Checklist
Infrastructure components Microservices hosts Configuration source Routing and/or autodiscovery engine Monitoring engine Logging engine
CQRS Synchronous queries Asynchronous commands send request and wait for response with data it’s acceptable to join data on a client side Asynchronous commands send request and wait for command unique id use command id to check execution status handling distributed transactions that affect several microservices can be tricky commands should be idempotent
CAP theorem Consistency Partitioning Availability
Eventual consistency After command for data modification was received, it may take some time for new data to be available for querying write entity Client app API DB id = 42 Client cache write [ 42 ]
Eventual consistency After command for data modification was received, it may take some time for new data to be available for querying read list Client app API DB [ 1, 20, 50, … ] Client cache read [ 42, 60 ]
Eventual consistency After command for data modification was received, it may take some time for new data to be available for querying read list Client app API DB [ 1, 20, 42, 50, … ] Client cache read [ 60 ] remove [ 42 ]
What about Azure?
Azure Service Fabric Service Fabric itself is platform agnostic Stateless and statefull services Service discovery Health monitoring Fault domains and upgrade domains Enterprise-ready https://docs.microsoft.com/en-us/azure/service-fabric/
Fault and upgrade domains Update 1 Node 1 Update 2 Node 2 Update 3 Node 3 Update 4 Node 4 Update 5 Node 5 Main Primary reserved Secondary reserved Microservice 1 1 2 3 Microservice 2 4 Microservice 3 5 Microservice 4 Microservice 5
Azure App Services .NET, Java, PHP, Node.JS, Python Shared web farm, isolated web apps Configuration management Backups Sticky sessions Built-in MySQL database https://docs.microsoft.com/en-us/azure/app-service-web/
Azure App Services NLB Server 1 CPU RAM Server 2 CPU RAM Server 3 CPU Distributed file system App 1 root App 2 root App 3 root App 4 root App 5 root
Azure App Services NLB Server 1 CPU RAM Server 2 CPU RAM Server 3 CPU Distributed file system App 1 root App 2 root App 3 root App 4 root App 5 root
Azure Functions Built on top of Azure App Services Small and independent unit of work Designed for handling huge amount of small parallel independent executions Consider them as event handlers Triggers and bindings Can be chained to handle sequential workflow Dedicated and consumption pricing plans https://docs.microsoft.com/en-us/azure/azure-functions/
Azure Functions
Azure Functions bindings Type Service Trigger Input Output Schedule Azure Functions ✔ HTTP Blob Storage Azure Storage Events Azure Event Hubs Queues Queues and topics Azure Service Bus Storage tables SQL tables Azure Mobile Apps No-SQL DB Azure DocumentDB Push Notifications Azure Notification Hubs Twilio SMS Text Twilio SendGrid email SendGrid
Azure functions quickstart Create cross-platform desktop apps using JavaScript and Azure https://www.youtube.com/playlist?list=PLi8gartTlTVx7sKg3XLuCQik1XSgUMNgl Sql Saturday 2017 Kyiv https://www.youtube.com/playlist?list=PLi8gartTlTVxR9lGYAXUNTk1Mo0qG1j7i
Questions? Anton Boyko Microsoft Azure MVP Ninja Azure Consultant boyko.ant@live.com