An expanded version of my previous presentation on DDD, this version was presented at the Sydney Architecture User Group, Thursday 27/05/2010, 6:00 PM, Grace Hotel , Kiralee or Pinaroo Function Room 77 York St Sydney, NSW. 2000

1. Jak Charlton jak@thinkddd.com www.thinkddd.com Readify www.readify.net

2. Practical Domain Driven Design Message Based Architecture and CQRS

3. Domain Driven Design IS An architectural methodology for evolving a software system that closely aligns to business requirements

4. Domain Driven Design IS NOT A silver bullet A panacea for all your troubles An easy path to follow Always the best solution And most importantly, it is not focused on the How, but the What and Why

5. Where Are We Going?

6. The Domain Vision Statement A shared understanding of what it is you are actually trying to create Should be brief, written in clear English and understood by business and tech people alike Should be factual, realistic, honest Should avoid superlatives and marketing speak Should avoid technical and implementation details

7. Domain = Sphere of Activity?

8. Domain A Domain is a Sphere of Knowledge, Influenceor Activity A Domain is represented by the Ubiquitous Language A Domain encapsulates a Domain Model A Domain lives within a Bounded Context

10. The Ubiquitous Language Human languages are Lossy Abstractions A major reason for failure of software projects is a failure of people, the failure to communicate The Ubiquitous Language is a shared language between the business and the development teams The UL comes from the business, and is enriched by the development teams

12. Domain Experts Domain Experts are the primary point of contact the development teams have with the business They are the Experts on their part of the business, not just users of the system They should have deep knowledge of the subject Domain

13. Looks Pretty Unique To Me

14. Entities Entities are the “things” within your Model An Entity is defined by being unique, and uniquely identifiable Entities have behaviour

15. You Can’t Tell One From Another

16. Value Objects Value Objects are the “things” within your model that have no uniqueness They are equal in all ways to another Value Object if all their properties match Value Objects are interchangeable

17. How Can We Make Sense?

18. Domain Model A Domain Model is a representation of the relationships between the Entities and Value Objects in your Domain It may look similar to UML or a class relationship diagram, but it is not one The Domain Model should be recognisable and understandable by the business

19. That’s a Lot of Bits

20. Aggregates “An aggregate is a collection of items that are gathered together to form a total quantity” - Wikipedia An Aggregate Root is the root item containing a number of parts that form a whole An AR is more likely to match a Use Case than any model structure

21. Everyone Loves an Event!

22. Domain Events The “3rd Thing” in DDD Like all events, notification that “something happened” They are in the past tense, and should be named accordingly Closely aligned to your Domain Model Will probably be handled by your messaging layer

24. Bounded Contexts When you have multiple models you should consider Bounded Contexts Each BC is a self contained “mini application” containing it’s own model, persistence and code base Within a BC, everything should be strictly consistent To map between BCs you use a Context Map

25. Transformer Domain Vehicle Domain

26. Context Maps A Context Map provides a clearly defined relationship between Bounded Contexts and the interactions between their Domain Models A Context Map may be represented in many forms, but it comprises the rules for how to turn A into B

28. Persistence Ignorance Subtitled “There Is No Database”  DDD uses the Repository pattern to create Persistence Ignorance A Repository represents itself as an in-memory list of items Repositories are specific to Aggregate Roots, not to Entities

30. Anti-Corruption Layers An Anti-Corruption Layer is a method to isolate two systems, allowing systems to be integrated without knowledge of each other An ACL presents a Facade to both systems, defined in terms of their specific models ACLs maintain the integrity of a Domain

32. Factors For Success of DDD Your domain is not trivial You have access to Domain Experts You have an iterative process You have a skilled and motivated team

33. Where Does DDD Work Best? Domain Driven Design is a way of dealing with complexity DDD is ideally suited to Behaviour Centric SystemsnotData Centric systems

34. Messaging Architectures DDD is ideally suited to Message Based Architectures DDD is ideally suited to Commands and Events DDD without Commands and Events is going to be a lot of hard work DDD is ideally suited to providing multiple autonomous systems that are loosely coupled

35. CAP Theorem Consistency: The client perceives that a set of operations has occurred all at once. Availability: Every operation must terminate in an intended response. Partition tolerance: Operations will complete, even if individual components are unavailable. You cannot have all three

36. ACID Atomicity, Consistent, Isolation, Durability Gives you Consistency and Availability Sacrifices Partition Tolerance Meaning: ACID systems are hard and expensive to scale

37. BASE Basically Available, Soft state, Eventually consistent Gives you Availability and Partition Tolerance Sacrifices Consistency Meaning BASE systems can scale easily and are very fault tolerant

38. BASE at a Database Level Most RDBMS databases are ACID Most NoSQL databases are BASE The largest expense in database systems is in fault tolerance and scaling BASE databases scale massively and cheaply BASE databases are highly fault tolerant

39. BASE at an Architectural Level Latency is the biggest constraint on architecture Unless you have pessimistic locking, all data is stale Most mistakes in developer code are around consistency vs. latency The main cause of rigidity and fragility in systems is due to trying to maintain consistency Command Query Responsibility Segregation gives us BASE at an architectural level

40. Command Query Responsibility Segregation (CQRS) Bertrand Meyer principle of CQS:every method should either be a command that performs an action, or a query that returns data to the caller At an architectural level this means:either issue commands, or issue queries, but never both And, query from a separate source from your domain commands

41. CQRS in a Picture Client Command Domain Event Handlers Queries (synchronous no bus) Event Handlers Publish Persist Update Domain Persistence Read Model

42. CQRS in a Simpler Picture Domain Read Model Events Commands DTOs Client

43. CQRS Useful Only for Scaling? CQRS gives us hugely simplified Domain persistence and leaves the Domain free of orthogonal concerns CQRS lets us focus on real Domain logic CQRS gives us optimised reporting and querying capability CQRS removes a large amount of the effort involved in dealing with inconsistent systems pretending to be consistent CQRS has real business benefits beyond scaling

44. Questions? Jak Charlton jak@thinkddd.com www.thinkddd.com Readify www.readify.net