September 2019 Summaries

Companies are increasingly migrating from Oracle to CockroachDB due to several factors, including the high Total Cost of Ownership associated with Oracle, the desire for cloud-based, distributed databases compatible with Kubernetes, and the need for flexibility to deploy on-premises and across various cloud platforms for global resiliency. The digital acceleration triggered by the pandemic has also pushed companies to seek modern solutions that surpass the limitations of legacy tech stacks. Cockroach Labs offers a comprehensive guide for migrating from Oracle to CockroachDB, acknowledging the challenges involved and providing a structured approach to extract, refactor, and load schemas and data, as well as refactor application SQL. Despite the complexity of the migration process, Cockroach Labs emphasizes its experience and readiness to assist customers through this transition, highlighting the advantages of adopting a cloud-native database with reduced complexity and scalability.

Bilal Akhtar's blog post chronicles his journey from being an intern to a full-time engineer at Cockroach Labs, detailing the unique aspects of his experience at the company. During his internships, Akhtar worked on projects critical to the product's success, such as adding SQL functionalities and optimizing the database, which provided him with significant learning opportunities and a sense of impact. He highlights the value of interning at smaller companies like Cockroach Labs, where he experienced a collaborative environment and gained insight into the broader company operations, contrasting this with the tendency to pursue internships at tech giants. Returning as a full-time engineer, he joined the Core Storage team, transitioning from his SQL-focused internships to work on the distributed systems that distinguish Cockroach Labs' database. Akhtar appreciates the company's emphasis on quality work and work-life balance, which significantly influenced his decision to return.

The blog post delves into the challenges of join ordering in SQL, a problem that has been complex since the early days of database management. It highlights the NP-hard nature of finding optimal join orderings, especially in general settings, and introduces the IKKBZ algorithm as a notable solution for specific cases. The algorithm, developed by Toshihide Ibaraki and Tiko Kameda, is rooted in operations research and leverages concepts from job scheduling problems, particularly those involving precedence constraints. The post explains how the IKKBZ algorithm applies when the query graph is a tree, and the solutions are left-deep without cross products, using nested-loop joins. It underscores the algorithm's significance as one of the few polynomial-time solutions with strong guarantees, although it is limited in application due to its specific requirements. Furthermore, IKKBZ serves as a useful starting point for more general heuristic algorithms, despite its constraints, and continues to hold relevance in the field of query optimization.