http://2017.springio.net/ - Caching Made Bootiful
Featuring Spring Boot, Hazelcast and JSR107 JCache.
Each of the modules starts listening to port 8080, Try with
http://localhost:8080/45
a few times to see how response time varies for the 45th result in the Fibonacci series.
A simple Spring Boot web service.
Add the @EnableCaching and @Cacheable annotations, and nothing else. Spring
will provide a default implementation of the caching manager, essentially a
concurrent hashmap.
This is great start, performance is great for the 2nd and subsequent calls. Until you restart.
This is a step on from version3, adding the hazelcast.xml file so Spring Boot
will create a Hazelcast server and use that for caching.
If you only run one such JVM, you won't notice much difference. If you run more than one, you can stop and restart either without losing the cached results.
A more honed approach to Hazelcast caching, using client-mode. This means the business logic -- the Fibonacci calculation -- is in the client, and the storage of the cached results is in the server(s).
Now we use JCache, JSR107, the Java caching standard.
Hazelcast is the implementation provider, and still in client-mode so the
same benefits of separating storage from processing as in version4.
But now it's standardised, so if you don't want to use Hazelcast you can swap in something dearer and slower.