Partition Recovery Patterns: A comparative analysis explained through illustration


When choosing persistent data stores for your distributed system, it is important to understand how particular products trade off consistency, availability, and partition tolerance. This talk will begin with a refresher of the CAP theorem and its relevance to architecting modern software systems, then proceed with a whistle-stop tour of 2-3 popular open-source data products (ie. RabbitMQ, Kafka, Cassandra — final line-up TBD), their comparative design priorities, and suggestions for how they can effectively be utilised for production workloads. All technical explanations will occur in the form of original sketchnote-style illustrations, with the possibility of some live-sketching!

Speaker

denise-yu

Denise Yu

 
Denise works at Pivotal Cloud Foundry, packaging open-source data projects such as RabbitMQ and Redis in wrappers (aka BOSH releases) so Cloud Foundry app developers can use them as backing data ...