Quarkus and Infinispan: winner combo
Infinispan 15.0 has been recently released, bringing along a wave of enhancements that Quarkus has also embraced. Let’s dive into this blog post to spotlight the significant updates and improvements for both Infinispan and Quarkus.
Serialization API improvements
Infinispan can handle various serialization formats, but Protobuf stands out as the most compatible one, offering full access to all Infinispan Cache features. It’s supported through the Protostream library. With Protostream 5, you not only get support for Protobuf 3 but also a more straightforward and user-friendly API.
@Proto
Using the @Proto annotation alone is sufficient to serialize a class, enum, or record with
Protostream. If you want to be more specific about what needs to be serialized, you can still
use the @Protofield annotation. However, a default schema setup will be automatically handled
for you with just a single annotation. This feature comes in handy, especially when your
schema undergoes frequent changes, and you need to refresh data frequently.
@Proto
public record Author(String name, String surname) {
}@ProtoSchema
The annotation used to define the schema has been renamed to @ProtoSchema.
So now, instead of @AutoProtoSchemaBuilder, you would use @ProtoSchema for specifying the schema.
This helps clarify its purpose and usage.
@ProtoSchema(includeClasses = { Book.class, Author.class }, schemaPackageName = "book_sample")
interface BookStoreSchema extends GeneratedSchema {
}Programmatic definition of schemas
If you need to define the schema programmatically, Protostream 5 now offers an API for that purpose. This allows you to define the schema dynamically in your code as needed.
@Produces
Schema magazineSchema() {
return new Schema.Builder("magazine.proto")
.packageName("magazine_sample")
.addMessage("Magazine")
.addField(Type.Scalar.STRING, "name", 1)
.addField(Type.Scalar.INT32, "publicationYear", 2)
.addField(Type.Scalar.INT32, "publicationMonth", 3)
.addRepeatedField(Type.Scalar.STRING, "stories", 4)
.build();
}Moreover, the MessageMarshaller class, which was previously planned for deprecation and removal,
has been reinstated and can continue to be safely used for implementing custom marshalling.
Support for mocks
Before Quarkus 3.9, because of the Search API, it wasn’t possible to mock
RemoteCache beans using @QuarkusTest and @QuarkusMock due to their Singleton scope.
However, starting from Quarkus 3.9, their scope has been changed to ApplicationScoped,
enabling full mocking for tests.
Additionally, there have been enhancements in the Search API.
Now, we can directly perform searches using methods exposed in the RemoteCache interface.
@Inject
@Remote("books")
RemoteCache<String, Book> booksCache; (1)
...
Query<Book> query = booksCache.query("from book_sample.Book b where b.authors.name like '%" + name + "%'"); (2)
List<Book> list = query.execute().list(); (3)| 1 | Books cache is injected |
| 2 | Use query method directly on the books cache |
| 3 | Retrieve the list as usual |
Infinispan Cache extension
Starting from Quarkus 3.11, the caching annotations in the Infinispan Cache extension have been deprecated. Infinispan now offers a new cache extension.
By swapping out the quarkus-cache dependency with quarkus-infinispan-cache,
you can utilize @CacheResult, @CacheInvalidate, and @CacheInvalidateAll annotations
from the quarkus-cache extension, while storing the data in the Infinispan Server.
@GET
@Path("/{country}/{city}/{name}")
@CacheResult(cacheName = "fruits")
public ExpensiveFruitsResponse getExpensiveFruitsResponse(@PathParam("country") @CacheKey String country,
@PathParam("city") @CacheKey String city,
@PathParam("name") @CacheKey String name,
@QueryParam("metadata") String metadata) {
invocations.incrementAndGet();
String id = UUID.randomUUID().toString();
String description = String.format("Fruit in city %s, with name %s", city, name);
return new ExpensiveFruitsResponse(id, description, metadata);
}Read all about it in the Infinispan Cache Extension Guide
Infinispan and Quarkus LangChain4j Integration
Now, you can use Infinispan as an Embedding Store in the Quarkus LangChain4j extension. Infinispan has full text features thanks to a special integration with Hibernate Search. Plus, starting from Hibernate Search 7.1, Vector Search is supported. This means you can use the Infinispan Server as an Embedding Storage for your LLM applications. Read all about it in the documentation.
Come Join Us
We value your feedback a lot so please report bugs, ask for improvements… Let’s build something great together!
If you are a Quarkus user or just curious, don’t be shy and join our welcoming community:
-
provide feedback on GitHub;
-
craft some code and push a PR;
-
discuss with us on Zulip and on the mailing list;
-
ask your questions on Stack Overflow.
