Skip to content

Roadmap

import { Badge } from ‘@astrojs/starlight/components’;

Cyntex v1 is divided into four release phases, each independently demo-able, runnable against real data, and covered by e2e tests.

Release Phases

POC

Goal: Skeleton is runnable; single-node CDC full pipeline works end-to-end

DeliverableStatus
Maven mono-repo skeleton (enforcer + ArchUnit + ${revision})✅ Done
Native CLI tooling spike (GraalVM 21, startup ~10ms)✅ Done
Full DSL grammar module (core-model / core-dsl / core-schema)🔄 In progress
Full connector catalog build pipeline (60+ connectors)🔄 In progress
Offline CLI (validate / new / explain, native binary)🔄 In progress
PDK loading + single-node basic CDC (MySQL → MongoDB)⏳ Not started
Control core (CRUD + lifecycle) + basic auth⏳ Not started

Exit criteria: CLI registers a connector + YAML single-table task (full snapshot / basic CDC) runs through + status query works


Alpha (~4 weeks, soft release)

Goal: FDM→MDM basic sync pipeline demoable; minimal AI integration

Deliverable
CDC connectors (directly included)
Basic processing nodes (rename / JS-GraalVM / typeFilter / unwind / date / row-level WHERE)
Minimal MCP (read-only + scaffold/explain, BYO-agent)
e2e test framework (declarative YAML spec + Java runner; built after DSL schema is locked)
OTel metrics / logging

Exit criteria: FDM→MDM sync task with basic processing operators runs end-to-end


Beta (~8 weeks, OSS core complete)

Goal: Distributed CDC + merge/join + orchestration + monitoring; CLI + Basic UI dual frontend

Deliverable
Hz distributed cluster (upgrade to 5.7.0 fork + cluster + quorum + heartbeat)
Distributed CDC + shared log mining (Ringbuffer multi-consumer) + offset persistence / resume
Cache management (Caffeine + Hz IMap L2 + CacheRegistry)
DuckDB join node
Master-detail merge operator (MERGE family)
DDL / Schema Drift handling
Multi-task dependency orchestration (task-of-tasks)
Complete MCP toolset (mutable CRUD + full lifecycle operations)
Basic UI (open source)
OpenLineage field-level lineage (introduced in Beta)

GA (~8 weeks, enterprise-ready)

Goal: External API + external data lake + automatic HA + closed-source enterprise plugin integration

Deliverable
apiserver (role=api) + API management — OSS open
Custom nodes
Debezium integration
Paimon incremental data lake (external storage, added at end of GA)
Automatic task HA (failover)
Containerization + port consolidation
Enterprise UI
Closed-source plugins (Tentatively Enterprise): RBAC / LDAP / Lineage & Tracing / enterprise connectors (DB2/Sybase/MSSQL/PG/Oracle agent)

Key Design Decisions

Locked invariants (see Architecture Decisions):

  • YAML-only authoring: SQL is removed as an authoring format; it appears only as an embedded value inside DuckDB nodes
  • BYO-agent, no bundled LLM: MCP server contains no model
  • No auto-fix: AI monitoring/alerting is read-only; mutable operations require human in the loop
  • Open core: No license gating; closed-source = SPI plugins
  • Single binary + --role: Merged deployment for small scale; role-separated for large scale

Current Blockers

ADRTopicImpact
ADR-0002Process merge form (Proposed, pending Accept)Gates full server-side POC
ADR-0017Testing strategy (Proposed, pending Accept)Gates e2e framework construction
ADR-0006PDK compatibility boundary (not written)Gates connector adapter layer