November 2025 Summaries

PlanetScale has introduced a new feature in its Insights platform called AI-powered index suggestions for PostgreSQL databases, aimed at enhancing query execution speed and overall database performance. By leveraging large language models (LLMs) and robust validation processes, the feature identifies optimal indexes by analyzing database workloads and suggesting new indexes only when necessary. The system addresses two main challenges: the LLMs' occasional inability to determine when changes are needed and their propensity for producing inaccurate results. To ensure reliability, the feature uses query performance data to filter queries that could benefit from an index and validates the suggestions through a two-step process, including syntactic checks and performance evaluation using the HypoPG extension. This ensures that only suggestions leading to substantial improvements are recommended, thereby optimizing database performance without unnecessary overhead.

PlanetScale has launched its $5 single node Postgres databases, offering a cost-effective solution for startups, side projects, and development environments without high availability (HA) features. These databases come with PlanetScale's developer-friendly tools, such as Query Insights, schema recommendations, and branching. The price for development branches has also been reduced to $5 per month. Users can easily scale their databases vertically by selecting different cluster sizes or switch to HA mode for added reliability. For those needing horizontal scaling in the future, PlanetScale is developing Neki, a sharded Postgres solution, ensuring that businesses can grow without the need to switch database providers. With PlanetScale's experience in managing some of the web's largest workloads, users can be confident in the platform's reliability and scalability.

Vitess 23.0.0 introduces enhancements designed to improve deployment, observability, and scalability of MySQL workloads, building on the foundation of version 22. Key updates include adopting MySQL 8.4 as a default, which ensures future compatibility and enhances operational tooling. New metrics offer deeper insights into transaction routing and shard behavior, while deprecated metrics and APIs are removed to streamline monitoring. Improvements in VTOrc and topology controls enhance operational safety for large-scale fleets, and updates to VTTablet and CLI/Docker workflows bring consistency and security. The release emphasizes better observability, operational strength, and future compatibility, with contributions from the Vitess community and PlanetScale team.

On October 20, 2025, an incident affected PlanetScale due to a DNS misconfiguration in a service provider, leading to capacity constraints and network instability within AWS us-east-1. The issue unfolded in two phases, initially impacting the PlanetScale control plane but leaving customer databases unaffected, followed by network partitions that affected some database branches. During the incident, PlanetScale engineers identified several challenges, including an inability to launch new EC2 instances and partial network partitions, which they managed with interventions such as changing default database regions and optimizing resource use. Despite the challenges, PlanetScale's design focus on isolation and static stability ensured minimal impact, with the resolution achieved through manual interventions and automated failover systems. Reflecting on the incident, PlanetScale plans to enhance resilience against the failure modes of dependent services and improve network partition tolerance, especially in the AWS us-east-1 region, which has six availability zones.

PlanetScale has introduced a more affordable entry point for its high-performance Metal databases, now offering plans starting at $50 per month, making them more accessible to budget-conscious startups. This new pricing model is available on both AWS and GCP platforms and allows customers to independently allocate CPU and memory from storage size, providing flexibility in meeting specific needs. The Metal databases have demonstrated significant performance enhancements, including reduced latency and increased queries per second, benefiting major clients like Cash App and Intercom. Previously, the $600 per month entry cost was prohibitive for some smaller businesses, but the new M-class clusters, which offer unlimited I/O and a range of configurations, aim to democratize access to these performance advantages. While smaller sizes for Postgres are already available, similar offerings for the Vitess system are pending due to the larger size of its fleet.