Why the project exists
Many high-availability architectures maintain additional database capacity for failover. That capacity is valuable, but it still carries server, storage, power, cooling, rack, licensing, monitoring, backup and maintenance cost. ShardronDB starts with a practical question: can more of the already-funded database infrastructure contribute during normal operation without turning every node into an uncontrolled writer?
The project does not reject active-passive HA, multi-primary replication, sharding middleware or distributed SQL. Each model is useful in the right context. ShardronDB explores a coordination layer above relational database nodes as another architecture option to evaluate.
Public design principles
- Gateway-controlled request validation and routing.
- Node-local ownership that makes the authoritative write location explicit.
- Backup-table placement and resilience planning across nodes.
- System-controlled primary key generation for normal client inserts.
- Separation of untrusted client traffic from trusted replication, migration and recovery flows.
- Monitoring and consistency checks as first-class architecture components.
- Progressive disclosure: public architecture first, protected runtime internals second.
Public and protected scope
Public
- architecture concepts;
- safe diagrams;
- database topic guides;
- evaluation questions;
- public Layer 1 repository;
- project updates and research context.
Protected
- runtime source code;
- replication engine internals;
- production SQL schemas;
- deployment credentials and secrets;
- customer data or database dumps;
- private operational runbooks.
Current project status
The first implementation work has focused on MariaDB-compatible relational environments. MySQL-compatible and PostgreSQL-compatible directions are part of the broader roadmap, but compatibility should be described only after engine-specific testing. Public materials therefore distinguish architectural intent from validated implementation evidence.
Before production adoption, the project needs reproducible tests covering correctness, conflict behavior, node failure, network partitions, backup restoration, rejoin, schema changes, performance and operational recovery. The Research & References page connects the project discussion with official database documentation.
Founder and contact
How to evaluate the project responsibly
- Start with a real workload and explicit RTO/RPO requirements.
- Identify current database ownership, replication and standby costs.
- List invariants that distributed writes must never violate.
- Define a limited proof-of-concept and measurable acceptance criteria.
- Compare the result with simpler alternatives.
- Do not infer production readiness from diagrams or marketing language alone.
Continue to the database architecture knowledge hub, public Layer 1 architecture, or evaluation request.