Zookeeper Repository
Using this library, and writing a few lines of code, you can manage your own domain objects in ZooKeeper. It provides CRUD operations and change notifications out of the box.
Who should use it?
There are many relational and NoSQL data stores you can use for persisting data, but the main purpose of ZooKeeper is for distributed coordination and configuration. So it makes sense to ask why should I use ZooKeeper as a persistence store? That's right. In most cases ZooKeeper is not a good choice for this reason, but for specific class of problems, we have found it useful, when you have a problem that met all of these conditions:
- You need online change notifications. Most relational and NoSQL databases do not provide a native push notification feature. One solution is to simulate push notification by periodic polls. But in ZooKeeper, you can simply listen for the changes.
- There are limited number of objects (i.e., thousands not billions). This library is supposed to handle all objects in memory. It even provides a method that gets the whole snapshot of objects in a Map data structure.
- You do not need facilities to check or force constraints like foreign keys. You can have multiple type of objects as different parallel repositories but you do not get automatic constraint checks about relations between objects.
Sample usage
As an example, suppose you have a simple data type (e.g., Car) and you want to persist it using ZkRepository. At first, you should extend your data type and implement two methods: a method that identifies the ID of a given object (that is used as node name in ZK), and a method that is used for validation of objects before persistence on add and update operations.
public class Car implements ZkRepositoryItem<Car> {
private int id;
private String model;
private String color;
...
@Override
int getId() {
return id;
}
@Override
public void validate(ZkRepository<Car> repository) {
if (isEmptyOrNull(model)) {
throw ValidationException("Car model is empty!");
}
}
}
Note that by default, a JSON representation of the objects will be stored in ZooKeeper. You can change the default serialization/deserialization mechanism by overriding serialize() and deserialize() methods.
After defining the car class, then implement the car repository by extending the general repository. You should call the super constructor by the ZK server addresses, and the root node in ZooKeeper in which the data objects should be kept:
public class CarRepository extends ZkRepository<Car> {
public CarRepository(String zkAddresses) {
super(zkAddresses, "apps_root/cars");
}
}
Now you are ready to use it:
// Initializing the repository: CarRepository carRepo = new CarRepository("zookeeper-server:2181"); carRepo.start(true); // Add operation: Car ferrari = new Car(1, "Ferrari", "Red"); Car ford = new Car(2, "Ford", "Blue"); carRepo.add(ferrari); carRepo.add(ford); // Get operation: assertEquals(ferrari, carRepo.get(ferrari.getId()); assertEquals(ford, carRepo.get(ford.getId()); // Update operation: ferrari.setColor("Gold"); carRepo.update(ferrari); assertEquals("Gold", carRepo.get(ferrari.getId()).getColor()); // Snapshot: Map<Integer, Car> snapshot = carRepo.getSnapshot(); assertEquals(ferrari, snapshot.get(ferrari.getId()); assertEquals(ford, snapshot.get(ford.getId()); // Remove operation: carRepo.remove(ferrari.getId()); Map<Integer, Car> snapshot = carRepo.getSnapshot(); assertNull(snapshot.get(ferrari.getId()); // Change notification: carRepo.registerChangeCallback( () -> System.out("Cars are changed. Current snapshot of values: %s", carRepo.getSnapshot().toString()));
8 Aug 4, 2022
8 Nov 8, 2022
7 Dec 21, 2022
14 Nov 8, 2022
2 Aug 4, 2022
442 Jan 3, 2023
8 Mar 15, 2022
9 Oct 15, 2022
1 Feb 3, 2022
182 Dec 26, 2022