About ReflexDB
Built to solve the problem every production database team eventually hits: reads that shouldn't be hitting your primary, but have nowhere else to go.
Why ReflexDB exists
Every application that grows past a certain point runs into the same problem. The reads that felt fine at launch — product listings, user profiles, search, dashboards — start competing with writes, migrations, and analytics queries on the same database. You add indexes. You tune queries. Then you consider a read replica, which solves the load problem but creates a new one: now you need to write an API layer on top of it.
ReflexDB is the read layer that exists between your database and your application. It syncs your schema in real time, generates a typed REST API from it, and deploys it on managed infrastructure. You don't write middleware. You don't manage a cache. You declare what you want to expose and ReflexDB handles the rest.
The core engine is written in C++ — not because we enjoy writing C++, but because query execution against a 200M-row in-memory snapshot needs to be fast and predictable. The whole system is designed to be operationally boring: provision, sync, query. No moving parts you need to babysit.
How we build
Correctness over speed
The engine has 172 tests across store, linker, query, types, server, and alias suites. The parser runs under libFuzzer with ASan and UBSan enabled. We'd rather ship slowly than ship something that silently returns wrong data.
Fully managed, nothing to operate
ReflexDB handles provisioning, scaling, and monitoring. You connect your source database, configure what to expose, and get a live API endpoint. If you terminate an instance, its resources are cleaned up immediately — nothing lingers.
No lock-in
The query API is plain REST. The config is a YAML file you own. The OpenAPI spec is available on every instance. If ReflexDB ever stops serving you, you can move to something else — your data is still in your source database where it always was.
Boring infrastructure
We run on standard cloud primitives — containers, queues, load balancers — not proprietary managed services or vendor-specific abstractions. Simple building blocks that are easy to reason about.
Where we run
Europe (Ireland)
Default region for all new instances. Hosted in the EU, GDPR-compliant by default. Your data does not leave the region.
US East (Virginia)
Available on all plans. Choose per instance at creation time, or set a default on your connection.
Data residency
Synced data is held in memory on dedicated compute instances in your chosen region. It is never stored persistently by ReflexDB, and never transmitted outside your region.
GDPR compliance
A Data Processing Agreement (DPA) is available on all paid plans. ReflexDB acts as a data processor; you remain the data controller for all customer data.
Get in touch
We're a small team. We read every email.
For support questions, the contact form is the fastest route. For everything else, email us directly.