Data Engineering Digest #15 (Aug 2020)
This edition came a little bit later than expected, but the highlights of this edition are still fresh and interesting.
To start, we have a new update of Kafka in their ongoing process to remove Zookeeper as a dependency, something that is planned to happen on version 3.0.
We also have the paper describing the inner workings of Delta Lake, published by Databricks. They give details on the motivations for Delta Lake, how they were able to achieve the ACID properties through an Object Store and Parquet files, while still enabling schema evolution, time travel, caching, among other really useful features.
And finally we have a benchmark comparing the latency of Apache Kafka and Apache Pulsar. In a nutshell, the results show that Pulsar has a more predictable latency than Kafka both on end-to-end latency as well as for the publish latency. The full description of the experiments and results can be found on the article.
New Tools & Updates
Data Engineering Role
Courses & Training
Publications
Delta Lake: High-Performance ACID Table Storage over Cloud Object Stores