Architecture of Network Services
The Resultity network consists of a set of modular backend services that manage job routing, node coordination, account control, and system observability. These services form the core of the centralized infrastructure — initially operated by the platform team, but designed to be progressively decentralized.
Each service is exposed via a dedicated subdomain and serves a distinct role in the system.
Core Services
| Service Name | Subdomain | Purpose |
|---|---|---|
| Cloud | cloud.resultity.com | API interface for inference requests (OpenAI-compatible). |
| Gate | gate.network.resultity.com | Account, node, subscription, and config management. |
| IO | io.network.resultity.com | Real-time job delivery to nodes via WebSocket or short HTTP. |
| Status | status.network.resultity.com | Cached metrics: active nodes, GPUs, models, and pricing. |
| Admin | admin.network.resultity.com | Internal interface for network control and emergency operations. |
| Space | space.resultity.com | Planned storage for agents, templates, RAG data, and automations. |
| Account | account.resultity.com | User dashboard for balance, keys, referrals, and account actions. |
These services communicate with shared storage systems and message queues, and serve both public and internal consumers.
Separation of Concerns
- Cloud Layer exposes public APIs and enforces key-based access and billing;
- Network Layer handles job assignment, node monitoring, and configuration updates;
- Account Layer manages user identity, node binding, credits, and referrals;
- Admin Layer performs diagnostics, emergency overrides, and global policy actions;
- Status Layer publishes real-time metrics and pricing data for transparency;
- Space Layer (planned) enables persistent user-deployed logic and assets.
Deployment and Redundancy
While initially centralized, each service is containerized and deployable as part of a horizontally scalable cluster:
- Replication allows traffic splitting and failover;
- Internal APIs (e.g., Gate to IO) are authenticated and rate-limited;
- Network updates are version-controlled and can be rolled back via snapshot triggers.
Future roadmap includes enabling validator-operated mirror services, reducing reliance on centralized endpoints.
Internal Responsibilities
Each service coordinates with internal queues and data stores:
- Pulling tasks from RabbitMQ;
- Storing and querying job metadata in MongoDB;
- Accessing node state from Redis;
- Calculating metrics and history in PostgreSQL.
These responsibilities are hidden from end users but critical to the system's orchestration logic.
The Resultity Network architecture provides the backbone for reliable, verifiable inference coordination — bridging decentralized execution with centralized scheduling, metrics, and billing control.