Foo really rocks. Example of using zio-http, zio-kafka, zio-schema, zio-rocksdb, zio-streams and a bunch of Scala libraries (especially zio ecosystem) to create an event sourcing application.
There are two entities: Stock and Movement. There are a few fixed stocks like amzn (Amazon),
nflx (Netflix) and glts (Glints).
case class Stock(
id: UUID = UUID.randomUUID(),
symbol: String,
price: BigDecimal = 0
);Movements are changes in price to the stock.
case class Movement(
stockId: UUID,
change: BigDecimal = 0
);So a movement with stockId for nflx with change=10 ie Movement("nflx-uuid", 10), is an event
which when consumed, changes the nflx stock by +10 ie nflxStock.price = nflxStock.price + 10.
The goal is to create an application that can receive these movements at high frequencies and update the stock correctly with low latency and good throughput.
Kafka is the event bus (or a distributed commit log) that receives, stores and shares Movement data
for stocks. It is also/will be used for changelog data to the Stock so when our application goes down
and comes back up, we start where we left off.
Technically not externally but more so embedded; This is the store we use to read and write stateful Stock
data. Changes to Stock after consuming Movement will be persisted via this db. RocksDB is used for persistence
and because it is embedded, it is fast. Not to mention the configurable block cache and the memtables.
This is a Kafka producer that generates random Movement data and produces it to Kafka.
Consumes Movement produced by the Simulator and changes the stock price, storing it in RocksDB
(and change logging into Kafka in the future)
zio-http application that is used to retrieve the stock data by id from the state store.
- docker
- docker-compose
- Java (preferably Java 17). If you are using M1 Mac, use SDKMAN and java 17.0.5-zulu
- sbt
docker-compose up -d broker
sbt run