June 2019 Summaries

You can now use stored procedures in YugabyteDB, which allow developers to "bundle up" operations that would otherwise take several queries and round trips into a single function, minimizing duplicate code. Stored procedures also help minimize transactions issues with functions, enabling developers to reuse existing stored procedures to perform the same actions. The latest release of YugabyteDB 1.2.11 includes stored procedure support, as well as 30 new features, enhancements, and bug fixes. This feature allows developers to enable a flag on their cluster to try out this functionality. To get started with stored procedures, developers need to enable the `ysql_beta_features` flag, create an accounts table, insert data into it, verify the data, create a stored procedure, call the stored procedure, and verify the transfer of money. The latest release also includes new documentation, blogs, and videos on various topics such as building Java, NodeJS, Go, Python, Ruby, and SQL Alchemy apps with ORM examples. Additionally, YugabyteDB will be attending several conferences and meetups over the next few months.

The GO-JEK team, in collaboration with YugabyteDB Engineering, recently presented the results of their benchmarking exercise on three cloud native databases - CockroachDB, TiDB, and YugabyteDB - on Kubernetes. The team used the Yahoo! Cloud Serving Benchmark (YCSB) dataset to evaluate the performance of these databases under various workloads. The results showed that YugabyteDB outperformed the other two databases in terms of throughput and latency for certain workloads, but had issues with running Workload E, which is focused on Short Range Scans. The GO-JEK team has been working to correct their misinterpretations and fix benchmarking-related issues, and the updated results will help end-users make informed choices when selecting a database.

This summary highlights the challenges of deploying stateful streaming applications on Kubernetes, specifically those using Confluent Kafka, KSQL, Spring Boot, and distributed SQL databases like YugabyteDB. These apps are inherently stateful due to their need to handle continuous data streams, which poses scalability, reliability, and functional depth challenges. To address these challenges, the article outlines four best practices for deploying streaming apps on Kubernetes: using StatefulSets for stateful apps, low-latency storage, high data resilience, and high-performance networking. The article also provides a reference architecture for deploying an IoT fleet management app on K8S, showcasing the deployment of Confluent Kafka, KSQL, Spring Boot, and YugabyteDB in a Kubernetes environment.

Data sharding is a solution for business applications with large data sets and scale needs, allowing them to distribute their data across multiple servers to improve scalability and performance. Sharding can alleviate the impact of unplanned outages by ensuring high availability, increase total cluster storage capacity, speed up processing, and offer higher availability at a lower cost than vertical scaling. However, manual sharding can complicate operational processes, require significant development complexity, and lead to uneven shard allocation, hotspots, and storing data on too few shards. Common sharding architectures include hash sharding, range sharding, and geo-partitioning, each with its own strengths and weaknesses. YugabyteDB is an auto-sharded distributed SQL database that supports both hash and range sharding and has support for geo-partitioning as a work-in-progress feature.

YugabyteDB has released version 1.2.10 with foreign key support, along with 16 new features and enhancements. This release includes improvements such as support for ALTER SEQUENCE syntax, HASH and RANGE keys, and DROP DATABASE syntax. Additionally, new documentation, blogs, tutorials, and videos have been added to help users get started with YugabyteDB. The company will be attending various conferences and meetups in the coming months, including PostgreSQL Meetups, SpringOne Tour, and AWS re:Invent. Users can share their experiences with YugabyteDB and receive a hoodie by telling the team about their use of the database.