Be the first to like this
Speaker: Tyler Denton, Sr. Solutions Developer, ClearCaptions
ClearCaptions is a telephone captioning service for people who are hard of hearing. Couchbase revolutionized the way ClearCaptions handles real time data streams. Processing an average of 150K minutes coming from 200+ concurrent calls handling 40,000+ calls per day, real-time data is challenging. CouchBase provides the speed, agility and reliability necessary to provide this service today by utilizing tools such as: N1QL, Map-Reduce Views, Spatial Views, Memory Optimized indexes, the CouchBase-Kafka Connector, and the simple CRUD commands available at the bucket level. We utilize CouchBase SDKs for Java, Python, and PHP and often use curl requests against the ReST API. At the heart of our platform is our Automated Call Distribution system, or ACD, a highly available cluster rooted in Java which interacts with CouchBase via CRUD commands and HTTP requests against our own microservice API. There are multiple user facing applications such as our agent station used to caption, real time dashboards and agent evaluators that interact with Couchbase via PHP, and Python using N1QL queries, Map-Reduce Views, and Spatial views. Integral to data sharing real-time is Kafka. We use the CouchBase-Kafka connector to replicate the CouchBase DCP stream to all our secondary data stores such as our reporting and sip user databases utilizing custom Python consumers to pull mutation events out of the Kafka topics linked to each Couchbase bucket. All of these components running together have given us an extremely strong, responsive, and fault-tolerant platform thanks to the use of Couchbase.