September 2018 Summaries

In this release of YugabyteDB, a new feature has been introduced that allows for document data modeling using the native JSON data type available in the YCQL API. This enables developers to benefit from both the structured query language of Cassandra and the document data modeling capabilities of MongoDB in a single database. The JSON data type is supported natively by YugabyteDB's DocDB storage engine, which makes it easy to search and retrieve attributes inside documents. The new feature allows for flexible schema management, fast data access, and efficient querying of attribute values using various operators such as `->>` and casting to integers or strings. This release provides a powerful tool for developers to build scalable and flexible applications that can handle complex data structures. With the introduction of JSON support, YugabyteDB is now more competitive with other NoSQL databases that offer native document data modeling capabilities.

In this latest release of 1.1, YugabyteDB introduces secondary indexes that can be used to speed up queries and enforce uniqueness of column values. Secondary indexes are fully decentralized with no single point of failure, making the cluster resilient to various faults. They also provide linear scalability, ACID compliance, high performance, and support for multiple indexes per table. The data for each secondary index is an internal table that is sharded into tablets, internally replicated, and distributed across nodes much like user tables. This allows for efficient point-queries when looking up by the index columns, and additional columns can be included to further increase performance.

YugabyteDB 1.1 offers various features including distributed ACID transactions, secondary indexes, and JSON document data types, which enable high performance applications with reduced costs and operational complexity. The database also boasts improved performance compared to Apache Cassandra, with some workloads showing a 2.5x speedup. Additionally, YugabyteDB supports Redis-compatible API enhancements, user authentication, and security features such as TLS encryption. The Enterprise Edition includes features like in-flight encryption, distributed backups, and read replicas, making it suitable for enterprise-grade deployments. With native support for Kubernetes and managed cloud providers, YugabyteDB 1.1 provides a scalable and flexible solution for developers and organizations.

Ravi Murthy has joined YugaByte as its VP of Engineering after spending almost 7 years at Facebook, where he led teams that managed the growth of applications and data. His experience with databases spans over two decades and includes stints at Oracle and Facebook. Ravi is excited to bring his expertise to YugaByte, a cloud-native distributed SQL database that aims to simplify business-critical OLTP apps by providing transactional consistency, high performance, and multi-region scalability. With its innovative approach, YugaByte has the potential to revolutionize data management for mission-critical enterprise applications, and Ravi is looking forward to being part of this journey as he scales the team and brings the product to the broader market.

YugabyteDB 1.1 is now officially available for download, offering improved YSQL capabilities, bug fixes, and performance enhancements. The new release also enhances the YEDIS API with features such as low-latency reads from followers and support for multiple databases or namespaces. In addition to these updates, YugabyteDB Enterprise has received orchestration integration with Pivotal Container Service (PKS) and Google Kubernetes Engine (GKE), as well as read replicas in beta. The company also welcomes Ravi Murthy, VP of Engineering, who brings experience from his previous roles at Oracle and Facebook. Recent customer wins include Xignite and CipherTrace, which will leverage YugabyteDB's features to improve their data architectures. YugaByte will be attending Pivotal's SpringOne conference next week and is running DIY Jepsen tests that have shown no correctness failures detected by the testing framework. The company has also stress-tested high-density database clusters with densities up to 26 TB across four nodes, demonstrating the performance and scalability of YugabyteDB. With these updates and recent hires, YugaByte aims to continue innovating and pushing the boundaries of distributed databases.

Jepsen is a testing framework designed to improve the safety of distributed databases, queues, consensus systems, and other similar systems. YugabyteDB, a robust, reliable, distributed OLTP database, has adopted Jepsen for testing its correctness and technical accuracy claims. The YugabyteDB team has performed their own DIY-style Jepsen testing, contributed enhancements back to the community, and are working on adding more tests to their suite. The Jepsen test framework subjects databases to various failure scenarios, such as node failures, network partitions, and clock skew, to verify that they live up to their marketing claims. YugabyteDB has been found to pass all its Jepsen tests with no correctness failures detected, and the team is working on adding more tests and engaging with Kyle Kingsbury for formal Jepsen testing. The test suite verifies atomicity, consistency, isolation, and durability in the face of various failure scenarios, and the YugabyteDB team has uncovered several issues while running these intense tests, including inconsistencies in read operations and linearizability inconsistencies.

YugabyteDB is an open source project that aims to build a high growth business while maintaining sustainability in the long run. The company chose a permissive Apache 2.0 license and an open self-governance model for its open source YugabyteDB project, allowing users to contribute back to the project through their own volition. To monetize the open source project, Yugabyte plans to adopt an open core business model with the intent of adding a managed service offering in the future. This approach aims to balance user adoption and commercial features while avoiding the existential challenges faced by companies that failed to monetize their open source offerings. The company also acknowledges the cloud provider threat and plans to compete by building high value products that users and customers love to use, similar to Elastic's approach. YugabyteDB is designed to be a transactional, high-performance, geo-distributed database with all critical features, making it an attractive alternative to other databases like CockroachDB, Google Cloud Spanner, MongoDB, and Azure Cosmos DB.