My first year with event-sourcing Froscon 25-08-2018
Slide 2
Tim Huijzers @Dragem Developer @ drukwerkdeal.nl Founder of DeventerPHP Deventer NL
Slide 3
First Try ● ● ● ●
Limited knowledge No experience No ES framework Doomed from the start
Slide 4
Second Try ● ● ● ● ● ●
Limited knowledge Limited experience New Framework New DI manager New ES Framework Doomed To Fail
Slide 5
Third Try ● ● ● ● ● ●
Some knowledge Some experience Known framework known database known ES framework Still doomed
Slide 6
Slide 7
Slide 8
BeerWarehouse
Slide 9
Why use Event Sourcing
Slide 10
CRUD
Slide 11
We will save a new entry in our system because we just bought it and will store it in the fridge for later.
Slide 12
If we change the location the system only knows about that location.
Slide 13
We drank it so it’s not in the system anymore
Slide 14
We want to keep a history of everything we drank.
Slide 15
I want to know when I drank this in my history.
Slide 16
But that’s only for new beers.
Slide 17
Slide 18
Events
Slide 19
Same Information as before + Explicit action about what happened
Slide 20
Make Small Events
Slide 21
Removed Location and changed name because in the real world you might not know this yet.
Slide 22
When returning home I put the beer in my fridge
Slide 23
I need room in my fridge so I take it out. Using the same Event
Slide 24
And at last a event about when I consumed it.
Slide 25
Slide 26
Crud ● ● ●
I know what beer I have. I know when it was consumed. I know where it is.
Event-Sourcing ● ● ● ● ● ● ● ● ●
I know what beer I have. I know when it was consumed. I know where it is. I know where it was before. I know when it was moved. I know where it was at any point in time I know how many times it was moved. I know when it was added to the system. I know what else was moved in that day.
Slide 27
“Every software program relates to some activity or interest of its user.” Eric Evans, Domain-Driven Design: Tackling Complexity in the Heart of Software
Slide 28
Slide 29
When To Use Event Sourcing
Slide 30
You need an audit log You like scalability You want to separate the read and write of an application You want to replay event on a dev machine to get an accurate situation of what the state was at a point in time. ● You want reporting but don’t know what yet. ● You are done with mapping objects to tables ● ● ● ●
Slide 31
When NOT To Use Event Sourcing
Slide 32
● ● ● ●
You only need a simple CRUD system. You are processing a lot of personal data. You just want to query a lot of things on the DB You are starting on a big project for production
Slide 33
Event Sourcing in code
Slide 34
Prooph
Slide 35
http://getprooph.org/
Slide 36
Slide 37
Command
Slide 38
Slide 39
Slide 40
Slide 41
Slide 42
Command Handler
Slide 43
Slide 44
Slide 45
Aggregate
Slide 46
Slide 47
Event
Slide 48
Slide 49
Slide 50
Slide 51
Slide 52
Slide 53
Slide 54
Back to the Aggregate
Slide 55
Slide 56
Slide 57
Slide 58
Slide 59
Think About Side Effects
Slide 60
Slide 61
Slide 62
Structuring your application
Slide 63
Slide 64
Understanding the DB
Slide 65
Slide 66
How many beers do I have?
Slide 67
How many different styles do I have?
Slide 68
How many beers have I drank last 30 days?
Slide 69
Projection
Slide 70
A Projection allows you to loop through all event (past and present) and build your own views. ●
Read Model ○
●
Projection ○
●
Define the data you would like to use. Loops through the events and applies that data to your view
Finder ○
Helps you find data from that view.
Slide 71
Slide 72
Slide 73
Pitfalls
Slide 74
Refactoring is harder, think about your architecture
Slide 75
Slide 76
Versioning ● ● ●
Change an Event but support the old version Make a new Event Make the Event right from the start
Slide 77
Something wrong with the event
Slide 78
Event are immutable, So don’t change them ● ● ● ● ●
Try solving it another way first. Correct errors with new events Try a upcaster Make a new stream and fill it with mutated events (and test) Change the events in the database
Slide 79
But what if I have like 100 trillion gazillion events?
Other Tools ●
Broadway ○ ○ ○
●
Axon ○ ○ ○
●
No Upcaster, No Snapshots, No Replaying Upcasting by MessageFactory, Snapshots by Trigger on event count, Replaying by Example code for replay
Akka ○ ○ ○
Upcasting by Event Adapter, Snapshots decided by actor, Replaying
Slide 94
Slide 95
Thanks, Any Questions? Example code from talk on: https://github.com/webbaard/BeerWarehouse