Build Predictive Machine Learning with Flink | Workshop on Dec 18 | Register Now
Venice is LinkedIn’s low latency derived database that powers most of the company’s AI use cases. Last year we open sourced Venice, which means that you (Yeah! You!!) can deploy and run Venice in your stack. Venice is designed from the ground up with asynchrony in mind, making it a great fit for ingesting the output of your stream processing jobs. It also supports Lambda and Kappa architectures out of the box.
This talk presents Venice and deep dives into how we designed it to enable high data ingestion volumes via Kafka, merging it all coherently from many data sources and many geographically distributed regions. Venice’s active-active replication allows write operations to be produced in any region and be consumed in any order by all other regions, while still achieving eventual consistency. We’ll cover how Venice’s conflict resolution strategy can be a powerful abstraction when dealing with many writers.