This project is a system consisting of 2 microservices. First microservice is a Client which accepts words through API and publishes them to message queue(Kafka). Second microservice(Processor) consumes messages from messaging queue and joins all words that are received in one minute to a sentence, then posts back this sentence to message queue. Client in its turn receives these sentences from the message queue and stores them in NoSQL Database(Cassandra). Also Client exposes service to perform a search.
TestTaskClient - folder with Client application.
TestTaskProcessor - folder with Processor application.
First, 2 Kafka topics should be created(each should have only ONE partition) - for words and sentences. Currently
they are called words_topic and sentences_topic but may be modified(see appropriate settings in application.properties of both apps). Address of Kafka brokers and Cassandra settings(Client app) are configurable as well.
To run both Client/Processorsapps, just move to their folders sequentially and execute in each one console command:
mvn spring-boot:run
After starting of both apps, Client endpoints are ready for use - they are described in file
resources/api-doc/swagger.yaml, it
may be opened in any suitable editor(e.g., http://editor.swagger.io). Comments in Client's ClientController.java
also contains necessary info.
Logging is organised with SLF4J, currently file and console outputs are used(see logback.xml).
Logic for sentences full-text search endpoint is based on SASI(https://docs.datastax.com/en/dse/5.1/cql/cql/cql_using/useSASIIndexConcept.html) but as it's warned that these indexes are experimental, makes sense to continue research - Elassandra (https://github.com/strapdata/elassandra) could be checked as alternative.
Application was tested on Kafka 2.0.0 and Cassandra 3.11.6.
If imagine what else could be done for further development of the project, health checks for both apps is a good point to consider. Apache Camel health checks mechanism (https://camel.apache.org/manual/latest/health-check.html), for example, could be taken into account.