Sugu Sougoumarane on Vitess for Postgres (Multigres) and Scaling Databases

Sugu's Sabbatical and Return to Tech

• sabbatical for three years, Sugu Sougoumarane returned to the tech industry due to a growing "itch" and the realization that infrastructure and databases still matter, even with the rise of AI.
• 💡 His sabbatical involved self-reflection, voluntary work, and advisory roles, leading him back to technical challenges.
• 🧠 The initial motivation for Vitess stemmed from addressing database scalability issues at YouTube, which experienced frequent outages.

The Genesis and Evolution of Vitess

• 🚀 Vitess began as a six-month project at YouTube to solve MySQL scalability problems, evolving into an intelligent proxy with connection pooling, query parsing, sharding, and routing.
• 💡 The project eventually became a fully distributed database solution, open-sourced and later forming the basis for PlanetScale.
• 📈 Vitess has seen significant adoption by major companies like Flipkart, HubSpot, Slack, and GitHub, with substantial contributions from the community.

The Need for Vitess for Postgres (Multigres)

• ⚠️ The Postgress ecosystem, despite its growth, lacked a robust, scalable solution comparable to Vitess for MySQL.
• 🎯 Sugu had previous attempts to bring Vitess to Postgres, but they were postponed due to commitments with PlanetScale.
• 💡 The "aha moment" for Vitess involved internalizing query analysis and sharding logic, moving beyond application-managed "keyspace IDs" to a system where the database proxy handles sharding computations.

Multigres: Technical Approach and Philosophy

• ✨ Multigres aims to be Postgres-native, leaving behind MySQL-specific features and legacy Vitess components.
• 🛠️ Key areas for improvement include enhancing high availability and consensus mechanisms within Postgres.
• 🧩 The project will leverage Vitess's learnings, including its pluggable sharding scheme concept and rigorous testing philosophy, while focusing on approachability over extreme flexibility.
• 🚀 The goal is to provide a solution that allows users to stay within the Postgres ecosystem even at massive scale, addressing limitations of current solutions like Neon and Aurora.

Building Multigres and the Future

• 🤝 Sugu joined Superbase due to their strong commitment to Postgres-native, open-source solutions.
• ⏳ An MVP is targeted within 3-6 months, with a focus on making the right architectural decisions for long-term iterative development.
• 💬 Sugu plans to publish blogs on high availability and consensus, emphasizing the importance of local storage and robust consensus for database performance and reliability.
• 💡 The ultimate aim is to bring back the era of screaming-fast databases with millisecond latencies, even at massive scales.