The document discusses Kudu, a new updatable columnar storage system for Hadoop that was built to address gaps in transactional and analytic capabilities of existing Hadoop storage technologies like HDFS and HBase. Kudu aims to provide both high throughput for large scans like HDFS and low latency for individual row lookups and updates like HBase, while supporting SQL queries and a relational data model. It leverages improvements in hardware by using a columnar format and indexes to improve CPU efficiency for these workloads compared to traditional storage systems. The document outlines Kudu's goals and capabilities and provides examples of use cases like time series analytics, machine data analytics and online reporting that would benefit from Kudu's simultaneous support for sequential