“Microservices” have become a trendy development strategy. Hosting and running such services used to be pretty painful... but here comes Service Fabric! Let’s take a closer look at this platform, its different development models and all the features it offers, and not only for microservices!
3. Agenda
• Microservices?
• Service Fabric?
• Microservices in Service Fabric?
Demo code on GitHub: https://github.com/n3wt0n
Slides on SlideShare: https://www.slideshare.net/n3wt0n
4.
5. What is a microservice?
• Encapsulates a single capability
• Are developed by a small engineering team
• Can be written in any language and framework
• Contain code plus state that is independently versioned, deployed, and scaled
• Interact with other microservices over well defined interfaces and protocols such as http
• Have a unique name (i.e. URL) that can be resolved
• Remains consistent and available in the presence of failures
• Ideal for DevOps automation!
6. Reasons to split a monolith into services
• Scaling independently (balancing cost with speed)
• Supporting different technology stacks
7. Reasons to split a monolith into services
• Two or more clients, where clients adopt new features at will
• Managing conflicting dependencies
8. Are microservices right for my app?
Maybe, if:
• Your application can be split in many different, independent pieces
• You need/want independent service scalability
• You need/want to have multiple technologies, multiple teams involved
• You are ready to change your DevOps practices and company culture
But:
• Any application that binds its components over a network connection introduces delays
• Security and Authentication get more complex
• Microservices are more complex to deploy, manage, and monitor
• You cannot assume that you can take an arbitrary system and break it into microservices.
Most systems acquire too many dependencies between their modules
9. From Monolith to Microservices
... we support any stage you choose
New or
transformed
microservices app
5
Parts of existing
monolith
extracted
4
Existing Monolith + new
microservices
3
Monolith Hosted as
guest executable or
container
2
Traditional app
1
5 stages in a continuum…
10. • Comparing an in-process call to an HTTP request
• Performance: Worse & increases network congestion
• Unpredictable time: Managing timeouts is challenging; pass deadlines
• Unreliable: Requires retry loops with exponential backup/circuit breakers
• Server code must be idempotent
• Security: Requires authentication, authorization, & encryption
• Debugging: network issues, perf counters/events/logs, causality/call stacks
IntelliSense, refactoring & compile-time type-safety)
13. Proven platform powering core Azure
and Microsoft services
Microsoft has deep expertise in running global
services such as Cortana, Skype & Cosmos DB
Service Fabric is the foundational technology
powering these services & core Azure infra
Sample scale of one of these services: 60 billion
events per day with millions of databases
Production scenarios with 3500 VMs
SQL
{ }Power BI
Dynamics
Intune
Cortana Skype
Cosmos DB
IoT Hub
Events Hub
SQL Database
14. Build: data-aware microservices
Lifecycle
Management
Always On
Availability
Orchestration Programming
Models
Health &
Monitoring
Dev & Ops
Tooling
Auto
Scaling
Programming
Models
Dev & Ops
Tooling
Orchestration Lifecycle
Management
Health &
Monitoring
Always On
Availability
Auto
Scaling
.NET or Java …
Built-in ASP.NET core integration;
work with VS and VSTS or
Eclipse and Jenkins
Reliable Services
Manage state reliability
without a database,
lowering latency
Guest Executables
Run existing code and
orchestrate life cycle using
service fabric
Reliable Actors
Implementation of actor design
pattern (Isolated, independent unit
of compute and state with single-
threaded execution)
Containers
Orchestrate your Windows
Server or Linux containers
reliably at scale
</>
15. Deploy: any code on any OS
Lifecycle
Management
Always On
Availability
Orchestration Programming
Models
Health &
Monitoring
Dev & Ops
Tooling
Auto
Scaling
Programming
Models
Dev & Ops
Tooling
Orchestration Lifecycle
Management
Health &
Monitoring
Always On
Availability
Auto
Scaling
Monitor and diagnose
Generate, aggregate, and analyze
events with built-in tooling and
integration with Azure services
Docker Compose
Orchestrate existing
container applications
natively
Automate
Deploy or remove applications
using PowerShell, CLI, Visual
Studio, and other APIs
CI/CD
Maximize uptime and scalability
with isolated compute threads
running concurrently
Rolling upgrades
Upgrade non-disruptively and
roll-back in case of failures,
automate with PowerShell
16. Operate: on any cloud at any scale
Lifecycle
Management
Always On
Availability
Orchestration Programming
Models
Health &
Monitoring
Dev & Ops
Tooling
Auto
Scaling
Programming
Models
Dev & Ops
Tooling
Orchestration Lifecycle
Management
Health &
Monitoring
Always On
Availability
Auto
Scaling
Scale programmatically
Use PowerShell, CLI, or APIs to
scale programmatically achieving
very high densities
Use controlled chaos
Test graceful and
ungraceful failure
scenarios
Recover gracefully
Recover from node or service
failure gracefully; replicate
data automatically
Use familiar tools
Such as Splunk, OMS, ELK, or
AppInsights to gain deep insights
or monitor application health
Secure at scale
Secure node-to-node
communication and user access
using built-in capabilities
17.
18. Awesome for developers
Use .NET, Java or more
• Stateless and Stateful services
• ASP.NET core integration
Visual studio integration
• Use single node cluster for dev/test purposes
• Use 5-node local cluster for extensive dev/test
Consistent experience across environments
• Run the same environment on dev box as in production
19. Awesome for dev-ops
Automated deployment
• Use PowerShell, Service Fabric CLI or APIs
Rolling upgrades for reliability
• Upgrade your services in an automated fashion
Visual studio team services integration
• For continuous integration and deployment
Integration with Jenkins and more
• Use tools familiar to you for deploying Java applications
20. Awesome for infrastructure
management
Cluster management
• Secure your clusters and handle failures gracefully
• Scale clusters programmatically or manually
• Ease maintenance with integrated OS patching
PowerShell, CLI, and GUI management
• Be productive and choose tools that suit your needs
Applications as an ARM resource
• Manage clusters as well as deploy app using templates
Health monitoring
• Monitor using integrated or 3rd party tool of your choice
• Query health using PowerShell or APIs programmatically
27. • All API requests must pass version info starting with v1
• New service versions must be backward compatible
• What about intra-service instance requests?
• During rolling update, old & new service instances run together
• Failure occurs if v2 instance makes v2 API request to v1 service instance
• Fix by performing a 2-phase update
1. Deploy v2 service instances (which accept v2 & v1 API requests)
• But never send v2 API requests
2. After all instances are v2, reconfigure instances to send v2 API requests
• OPTIONAL: After all instances upgraded, deploy again with v1 code removed
28.
29.
30. • Services start/stop, scale up/down, & crash/move
• Each instance has its own endpoint (dynamically-assigned IP/port)
• Clients must dynamically discover service instance endpoints and load balance between
them
• Beware: a new service instance could be assigned a previous instance’s endpoint
• This requires certificates or some ID/uniqueness so client knows which
service it’s communicating with
37. Stateless microservices pattern
• Stateless services can have an external endpoint
• Configure the endpoint in ServiceManifest.xml (especially the port)
• For external accessible services, remember to configure Load Balancer
• Stateless service instances are equivalent and activated “randomly”
40. Stateful microservices pattern
• Stateful services cannot have an external endpoint
• Stateful services must specify partitions
• Stateful service instances are not equivalent and not activated “randomly”
• Stateful services use Reliable Collections to persist the data
41. Reliable collections
• Reliable collections make it easy to build stateful services.
• An evolution of .NET collections for the cloud.
Collections
• Single machine
• Single threaded
Concurrent
collections
• Single machine
• Multi threaded
Reliable collections
• Multi machine
• Replicated (HA)
• Persistence (durable)
• Asynchronous
• Transactional
Scale stateless services backed by partitioned storage
Increase reliability and ordering with queues
Reduce read latency with caches
Manage your own transactions for state consistency
Stateful services eliminate much of the latency and complexity in applications that uses the stateless pattern and are unique to Service Fabric.
Application state resides in the compute tier
Low latency reads and writes
Partitions are first class at the service layer for scale-out
Built in transactions
External stores for exhaust and offline analytics