How Atlas scales hundreds of merchant databases with Cloud SQL Enterprise Plus edition

Atlas, a restaurant operating system provider, migrated hundreds of merchant databases to Cloud SQL Enterprise Plus edition, reducing database operations time by 30% and enabling 200-300% annual growth. The migration eliminated separate connection pooling, provided query insights for performance tuning, and allowed near-zero downtime scaling.

Atlas https://www.atlas.kitchen/ is building the operating system for restaurants. Online storefronts, point of sale, third-party logistics, food platform integrations, customer loyalty, and AI tools represent everything a restaurant needs to start, run, and grow. We work with brands like SaladStop, Killiney, Haidilao, Raffles Hotel, Lo and Behold Group and the Les Amis Group in Singapore, helping merchants increase basket sizes, grow sales, and reduce operational costs. Every merchant on Atlas gets their own dedicated Cloud SQL for PostgreSQL https://cloud.google.com/sql/postgresql database. Restaurants are very different from each other. A single-outlet cafe and a multi-outlet chain should not look the same underneath. Isolated databases give us full data separation, predictable performance even during peak lunch and dinner rushes, and the flexibility to scale, tune, or migrate each merchant independently. As Atlas grows, the number of databases grows with us. We started on the standard Cloud SQL Enterprise edition. It was a solid foundation, but as we onboarded more merchants and shipped more features, the operational layer required to manage our databases became a bottleneck. We were managing connection pooling as a separate layer, which meant more services to run, secure, and monitor. When a query caused a CPU spike, we needed to know exactly what happened and which merchant triggered it, but we were spending too much time reconstructing problems from limited signals. With a lean team and no dedicated database engineers, every extra component multiplied the maintenance load. When we needed to provision new database instances, the Google Cloud team introduced us to Cloud SQL Enterprise Plus edition. We were already asking ourselves how much more operational overhead this was going to add, and what stood out was that Enterprise Plus edition removed whole categories of work we would otherwise have to own. Managed connection pooling: Now built directly into Cloud SQL, we no longer run pooling as a separate layer. This means fewer moving parts, less to maintain, and a smaller security surface area. Query insights: This was the most impactful feature for our needs. We can now see exactly which queries are expensive and which merchant is triggering them. It turns performance tuning from guesswork into something concrete and actionable. For a platform running hundreds of databases, this visibility is a "superpower." Data cache: This keeps read performance consistent even as merchant datasets grow. Since restaurants generate more data every day, the data layer needs to stay fast as that complexity compounds. Near-zero downtime scaling: We can now scale instances as merchants grow without disrupting service during off-peak hours. After seeing the results on the new instance, we migrated all our existing databases to Enterprise Plus edition as well. Atlas today powers thousands of restaurant outlets, processes tens of thousands orders daily using hundreds of managed databases. The biggest change is where engineering time goes. We spend 30% less time on database operations and more time building products. Merchant onboarding got simpler because a new merchant is provisioned in seconds with a ready-to-use managed database. We are much more proactive on performance now, catching and fixing issues before they reach merchants. Day to day, we are not thinking about database plumbing. We are thinking about how to serve merchants better and that has allowed Atlas to grow 200% to 300% year over year. We are investing deeply in AI, both internally and externally. Internally, we have gone all in on agentic engineering through AI-assisted development workflows that let a lean team build, review, and ship code significantly faster. Externally, we are building AI-powered tools that help restaurant operators make better decisions and act on them. We have a lot of experimental ideas on the roadmap, including new product surfaces and new ways to help restaurants grow. The thing that gives us confidence to move fast on all of this is that the foundational layer, Cloud SQL and Google Kubernetes Engine https://cloud.google.com/kubernetes-engine GKE , is battle-tested and does not get in the way. Google Cloud handles the infrastructure complexity. Atlas stays focused on building the best tools for restaurants. Cloud SQL Enterprise Plus gave us a database architecture that is flexible, observable, and easy to scale. We are not thinking about infrastructure anymore, we are thinking about our merchants. As we go deeper on AI and continue growing the platform, Google Cloud gives us the confidence to move fast without worrying about what is underneath. Don't let infrastructure bottlenecks slow down your innovation. Whether you are managing tens or hundreds of databases, see how Google Cloud SQL can streamline your operations, enhance observability, and give your engineering team the freedom to focus on what matters most.