All docs
Docs
Getting Started

Production single-node architecture

Synapsor Cloud is controlled beta today. This page separates the current hosted beta, the production-grade single-node architecture being hardened, and future distributed/serverless work.

Overview

Synapsor Cloud V2 is controlled beta today. Do not read this page as a claim that the hosted service is already a general production-grade, HA, serverless, compliance-attested database.

The production-grade single-node milestone is narrower and honest: one managed single-node runtime per durable database boundary, verified backups and restore drills, encryption, audit logs, stable API errors, clear RPO/RTO targets, operator runbooks, tenant quotas, billing gates, and no misleading HA or serverless claims.

For existing Postgres/MySQL apps, the production-oriented path keeps the existing database as source of truth. Synapsor live-reads selected rows through least-privilege mappings, stores the agent contract, evidence, audit, proposals, settlement, replay, and writeback status, then lets a trusted worker apply approved changes through the app's normal database rules.

CDC/mirrored subsets are private preview/hardening work, not the current public connector path. They exist as preview/dev-test primitives in the codebase, but production CDC requires replication-slot/binlog lifecycle management, lag/backpressure monitoring, retention controls, restore semantics, fail-closed operations, and operator signoff before customer production use.

Future distributed/serverless Synapsor is a separate architecture track. It requires remote WAL/page or segment services, durable consensus/replication, compute-storage separation, branch-aware indexes at scale, and failover mechanics. The single-node production milestone does not claim those properties.

Three launch tracks, three different claims

The controlled beta is available now. Production single-node is being hardened. Distributed/serverless is future architecture work.

Controlled beta today
  • Hosted single-node beta
  • Free invite approval
  • Builder Trial / paid Builder
  • No SLA/HA/compliance claim
  • Operator-controlled restore/PITR
Production single-node target
  • Verified backups and restore drills
  • Published single-node RPO/RTO
  • Tenant quotas and billing gates
  • Audited operator actions
  • Production support and incident runbooks

Developer notes

  • Use controlled beta for evaluation and trusted early workflows, not as a blanket production-grade claim.
  • Before production-data support, require restore verification, incident runbooks, support process, security review, operator MFA, key rotation, billing gates, and a clear RPO/RTO.
  • Use live-read external mappings for current existing-DB adoption.
  • Do not market CDC as production-ready until lag, backpressure, retention, cleanup, restore, and fail-closed behavior are operationally proven.
  • Do not claim HA, serverless, horizontal scaling, compliance certification, or SLA until those systems and reviews exist.