This document provides an overview of architecting microservices using .NET. It discusses why microservices are used, common architecture patterns, and implementation considerations. Key points include:
- Independent, loosely coupled services that are fault tolerant and easy to scale are goals of a microservices architecture.
- Communication between services should be kept simple, using either synchronous HTTP or asynchronous messaging. Synchronous calls can lead to temporal coupling so circuit breakers and failure handling are important.
- Domain-driven design principles like bounded contexts and separating queries from commands (CQRS) can help define appropriate service boundaries and responsibilities.
- Event sourcing avoids shared state and two-phase commits by persisting a sequence of events rather than
3. Preamble
As an industry we’re still learning and maturing
Tooling for microservices is akin to frameworks
for JavaScript
Many microservice implementations aren’t
‘pure’… and that’s OK
Docker onWindows is still a work in progress
Out of scope: Azure Service Fabric,Akka.NET,
etc.
Lots of bullet points so you can use this deck as
a reference
6. OK. More
seriously…
Greater flexibility & scalability
More readily and easily evolvable systems
Independently deployable parts
Improved technical agility
Independent development teams
7. A few more
reasons?
Sure!
Resilience.A failure in one service shouldn’t
wipe out the whole system.
Tech stack flexibility. Right tool for the right
job.
Smaller services are easier to understand and
maintain.
A potential migration approach for legacy
systems
9. WTF!OMG!
Gah! <update
resume />
Isn’t this meant to be easy?!
I can’t tell how it fits together anymore!
It’s more brittle now than it ever was!
Performance is terrible!!
I still need to make code changes in multiple
services at once!Yyyy?!
10. Why?
Distributed systems are hard
Eventual consistency messes with your head
Old habits result in a distributed “big ball of
mud”
Challenges debugging across multiple services
People tend to forget good design practices
through underuse (legacy habits)
11. Other things
to deal with
Shared databases across services?
Do transactions across services mean we need 2
phase commits?
What is the current “version” of the system?
How do we do authentication and authorization
at a service level?
What do testers actually, you know… test?
13. A quick
reminder
Architecture isn’t about technology alone.
Think of your teams and their skills.
Consider your organisation’s structure.
Keep your architecture simple so you can meet
your customer’s needs in the most cost effective
way.
Don’t build what you don’t need.
Don’t build what you might need!
14. Key goals
Independent, loosely coupled services
Cheap to replace, easy to scale
Fault tolerant, version tolerant services
18. KeepComms
Simple
Use language & platform agnostic
communications
One synchronous approach
(e.g. JSON over HTTP)
One asynchronous approach
(e.g. AMQP using RabbitMq)
Consistency in comms reduces complexity.
19. Synchronous
Comms infers
Temporal
Coupling
If your services use synchronous comms, you
need to handle failures and timeouts.
Use a circuit breaker pattern
Design with failures in mind (and test it!)
Netflix created the “chaos monkey” for testing
failures in production.
http://www.lybecker.com/blog/2013/08/07/automatic-retry-and-circuit-breaker-made-easy/
20. Identify your
business
transactions
One user request may result in tens, or
hundreds, of microservice calls
Treat each user request as a logical business
transaction
Add a correlation ID to every user/UI request
Aids with request tracing and performance
optimisation
Aids with debugging, failure diagnosis and
recovery
21. Evolvable
APIs via
Consumer
Driven
Contracts
A concept from the SOA days:
“Services aren't really loosely coupled if all parties
to a piece of service functionality must change at
the same time.”
In SOA daysWSDLs and XSDs were meant to
solve this.Yeah… right.
With HTTP APIs, have a look at Pact
http://www.infoq.com/articles/consumer-driven-contracts
https://github.com/SEEK-Jobs/pact-net
https://www.youtube.com/watch?v=SMadH_ALLII
22. Loose
Coupling and
Service
Discovery
Loose coupling implies no hard coded URLs.
Service discovery isn’t new (remember UDDI?)
Microservices need a discovery mechanism.
E.g. Consul.io & Microphone
https://github.com/rogeralsing/Microphone
25. Design
Patterns &
Components
Domain Driven Design – Align services to
Domain Contexts, Aggregates & Services
CQRS – Command Query Responsibility
Separation. Scale reads and writes
independently.
SQL/NoSQL – Persistent, easily rebuilt caches
for query services.
Versioning – APIs are your contracts, not
versions of binaries.
26. Design
Patterns &
Components
Message Bus – Reliable, async comms.
Optimistic Concurrency – No locking!
Event Sourcing – Persist events, not state.Avoid
2-PC hassles.
Application Services – encapsulate access to
microservices; optimise for client needs.
27. Event
Sourcing.
Really?
Yes, really :-)
When a domain object is updated, we need to
communicate that change to all the other
interested microservices.
We could use 2-phase commit, and we could
also drink battery acid.
With ES we simply save a domain event to our
event store and then publish it on the bus.
Interested other services subscribe to events
and take action as they see fit.
28. WARNINGS
The “100 line” rule is just silly.
These nano-services are effectively a service-
per-method.
Turn your app into thousands of RPC calls!Yay!
Services should be “business services” and
provide business value.Again, apply DDD
concepts to determine service boundaries.
29. Rule of
thumb for
sizing
microservices
Have a single purpose
E.g. manage state of domain entities
E.g. send emails
E.g. authenticate users
Be unaware of other services (in the core)
Consider use case boundaries/bounded
contexts
32. Web API Controller
Request (HTTP)
Aggregate
Event Handler(s)
Event Store
Domain Micro Service
Command
Message Bus (publish)
Command Handler
Command(s)
Event Store Repository
Save New Events
Event(s) Event(s)
33. Web API Controller
Query (HTTP)
Query Handler
Event Handler(s)
Message Bus (subscribe)
Query Micro Service
Event(s)
Read Model
Persistence
(akaView Store)
Consider splitting
here when scaling
beyond a single
instance to avoid
competing
consumers
Query
Updates
38. Boundaries?
User Story?
As the coffee shop owner
I want to define the products
that are offered for sale
So I can set my menu
Use Case?
Manage Products
View products
Create/Update products
40. SceneSetting
Domain entities are in the application core and
updated via methods
Commands/Queries are the adapters and ports
for our services
CQRS – use separate microservices for
commands and queries
42. Event
Sourcing
impacts
design
As Event Sourcing is used, domain objects
ONLY update their state by processing an
event.
Commands do not update state.
Commands cause events to fire.
Useful when replaying events.
43. public class Product : Aggregate
{
private Product() { }
public Product(Guid id, string name, string description, decimal price)
{
ValidateName(name);
ApplyEvent(new ProductCreated(id, name, description, price));
}
public string Name { get; private set; }
public string Description { get; private set; }
public decimal Price { get; set; }
private void Apply(ProductCreated e)
{
Id = e.Id;
Name = e.Name;
Description = e.Description;
Price = e.Price;
}
44. Aggregate
BaseClass
Holds unsaved events
Helper method to reapply events when
rehydrating an object from an event stream
Provides a helper method to apply an event of
any type and increment the entity’s version
property
45. public abstract class Aggregate
{
public void LoadStateFromHistory(IEnumerable<Event> history)
{
foreach (var e in history) ApplyEvent(e, false);
}
protected internal void ApplyEvent(Event @event)
{
ApplyEvent(@event, true);
}
protected virtual void ApplyEvent(Event @event, bool isNew)
{
this.AsDynamic().Apply(@event);
if (isNew)
{
@event.Version = ++Version;
events.Add(@event);
}
else
{
Version = @event.Version;
}
}
46. public class Product : Aggregate
{
private void Apply(ProductNameChanged e)
{
Name = e.NewName;
}
public void ChangeName(string newName, int originalVersion)
{
ValidateName(newName);
ValidateVersion(originalVersion);
ApplyEvent(new ProductNameChanged(Id, newName));
}
47. Port:
Command
Handlers
Commands do not have to map 1:1 to our
internal methods.
Commands (the ports) represent the inbound
contract our consumers rely on.
Internal implementation and any domain
events are up to us.
Command objects are just property bags.
48. public class ProductCommandHandlers
{
private readonly IRepository repository;
public ProductCommandHandlers(IRepository repository)
{
this.repository = repository;
}
public void Handle(CreateProduct message)
{
var product = new Products.Domain.Product(message.Id, message.Name,
message.Description, message.Price);
repository.Save(product);
}
50. [HttpPost]
public IHttpActionResult Post(CreateProductCommand cmd)
{
if (string.IsNullOrWhiteSpace(cmd.Name))
{
var response = new HttpResponseMessage(HttpStatusCode.Forbidden) { //… }
throw new HttpResponseException(response);
}
try
{
var command = new CreateProduct(Guid.NewGuid(), cmd.Name, cmd.Description, cmd.Price);
handler.Handle(command);
var link = new Uri(string.Format("http://localhost:8181/api/products/{0}", command.Id));
return Created<CreateProduct>(link, command);
}
catch (AggregateNotFoundException) { return NotFound(); }
catch (AggregateDeletedException) { return Conflict(); }
}
51. Adapters:
Database &
Message Bus
Repository pattern to encapsulate data access
Event sourcing; persist events not state.
Immediately publish an event on the bus
Note:This approach may fail to publish an event
Can be prevented by using Event Store as a pub/sub
mechanism
Can also be prevented by publishing to the bus and using a
separate microservice to subscribe to and persist events to
the EventStore
Personal choice: RabbitMq for ease of use & HA/clustering
52. public async Task SaveAsync<TAggregate>(TAggregate aggregate) where TAggregate : Aggregate
{
//...
var streamName = AggregateIdToStreamName(aggregate.GetType(), aggregate.Id);
var eventsToPublish = aggregate.GetUncommittedEvents();
//...
if (eventsToSave.Count < WritePageSize)
{
await eventStoreConnection.AppendToStreamAsync(streamName, expectedVersion, eventsToSave);
}
else { //... multiple writes to event store, in a transaction }
if (bus != null)
{
foreach (var e in eventsToPublish) { bus.Publish(e); }
}
aggregate.MarkEventsAsCommitted();
}
56. var eventMappings = new EventHandlerDiscovery().Scan(productView).Handlers;
var subscriptionName = "admin_readmodel";
var topicFilter1 = "Admin.Common.Events";
var b = RabbitHutch.CreateBus("host=localhost");
b.Subscribe<PublishedMessage>(subscriptionName, m =>
{
Aggregate handler;
var messageType = Type.GetType(m.MessageTypeName);
var handlerFound = eventMappings.TryGetValue(messageType, out handler);
if (handlerFound)
{
var @event = JsonConvert.DeserializeObject(m.SerialisedMessage, messageType);
handler.AsDynamic().ApplyEvent(@event, ((Event)@event).Version);
}
},
q => q.WithTopic(topicFilter1));
60. Redis
Repository
Redis: A key/value store, with fries
Collections stored as ‘sets’
Convention approach to ease implementation
Single objects stored using FQ type name
Key = MyApp.TypeName:ID
Value = JSON serialised object
All keys stored in a set, named using FQTN
Key = MyApp.TypeNameSet
Values = MyApp.TypeName:ID1, MyApp.TypeName:ID2, etc
Redis can dereference keys in a Set, avoiding
N+1 queries.
61. public IEnumerable<T> GetAll()
{
var get = new RedisValue[] { InstanceName() + "*" };
var result = database.SortAsync(SetName(), sortType: SortType.Alphabetic, by: "nosort", get: get).Result;
var readObjects = result.Select(v => JsonConvert.DeserializeObject<T>(v)).AsEnumerable();
return readObjects;
}
public void Insert(T t)
{
var serialised = JsonConvert.SerializeObject(t);
var key = Key(t.Id);
var transaction = database.CreateTransaction();
transaction.StringSetAsync(key, serialised);
transaction.SetAddAsync(SetName(), t.Id.ToString("N"));
var committed = transaction.ExecuteAsync().Result;
if (!committed)
{
throw new ApplicationException("transaction failed. Now what?");
}
}
64. Hey, Mister!
I don’t want
your Docker
Kool-Aid!
That’s cool.You don’t need Docker (or
containers).
Always get the latest code you need.
Then manually build & run all of the services on
your dev box each time you test.
Use scripting to make it a little less painful.
Side-effect: Encourages a low number of
services.
65. Use service
subsets to
ease the pain
Use test/mock services.
Only spin up the services you need to test your
work, and avoid all the other services that exist
Requires a bit more knowledge around what
services to use and what to mock.
Could also use tools like wiremock to intercept
and respond to HTTP requests.
66. Production in
a box
Use Docker images and Docker-Compose to
automatically build and run environments that
match production.
You may be limited by the resources of your dev
box (RAM, CPU cores, disk)
Could also use Azure RM templates or Azure
Container Services to spin up environments in
the cloud. (or the equivalent in AWS)
67. Containers
and
versioning
Don’t think about “upgrading” microservices.
Containers are immutable.
You don’t upgrade them; you replace them.
No more binary promotions to prod.
You promote containers to prod.
Have an image repository (e.g. artifactory)
68. What is the
version of the
app?
Consumer Driven Contracts reduce the care
factor somewhat.
Consider having an environment configuration
file
List the version of each microservice that has been
tested as part of a “known good” configuration
-- OR --
Ignore versioning and rely on monitoring in
production to report problems, and rollback
changes quickly