Skip to content

Roadmap and project status

ElastiCluster is built in thirteen milestones, from foundations to v1.0. Two are done, the third is being built, and nothing has been released yet. This page says exactly what works today and what each release is planned to add.
  1. M0, Foundations: done
  2. M1, Servers and agents: done
  3. M2, Plans, approvals, projects and environments: in progress
  4. M3, PostgreSQL high availability: planned for v0.1
  5. M4, Backups and restore: planned for v0.2
  6. M5, Monitoring, alerts and DBA tools: planned for v0.3
  7. M6, MCP and AI operations: planned for v0.3
  8. M7, MongoDB, Valkey and Dragonfly: planned for v0.4
  9. M8, MariaDB Galera: planned for v0.5
  10. M9, Garage S3 object storage: planned for v0.6
  11. M10, PostgreSQL day-2 completeness: planned for v0.7
  12. M11, Platform and fleet operations: planned for v0.8
  13. M12, Hardening and general availability: planned for v1.0

Status as of 2026-10-14.

Working in development builds

These parts are implemented, tested and committed. They ship in the first release.

  • The control-plane server, its REST API with OpenAPI and stable error codes, and live updates.
  • A setup wizard with a one-time setup token, the owner account, two-factor codes and recovery codes.
  • Sign-in with two-factor codes, sudo re-authentication, roles and personal access tokens.
  • An encrypted secrets vault with master-key rotation, and an internal certificate authority.
  • A hash-chained audit log with signed anchors, verification and export.
  • Adding servers over SSH or with a one-line install and join token, with admin approval.
  • Preflight checks with one-click remediations, and host preparation.
  • The agent in container or binary mode, with signed self-updates that restart no managed container.
  • Declarative plans for servers that repair drift, and the data identity guard.

Being built now

Milestone 2 brings the machinery every later feature uses:

  • Plans with diffs, risk and disruption; policies and approvals with four-eyes rules; stale approvals and re-plan.
  • Durable operations that pause on failure and resume after restarts; maintenance windows.
  • Projects and environments with policies, quotas and an egress allowlist; invitations.
  • Self-snapshots of the control plane and a guarded restore mode.
  • The first ecl CLI commands and the configuration generators for HAProxy and keepalived.

These are proven on a test-only engine first. Real databases start with PostgreSQL in the v0.1 preview.

Milestones and releases

Milestones end in a tagged pre-release when they add something you can use. There are no dates: a milestone is finished when its exit tests pass, not when a calendar says so.

Milestone Theme What it adds Ships in State
M0 Foundations Control-plane server, setup wizard, sign-in with two-factor codes, roles, encrypted secrets, internal certificate authority, tamper-evident audit log, live updates. v0.1 Done
M1 Servers and agents Adding servers over SSH or with a join token, admin approval, preflight checks, the agent as a container or binary, signed self-updates, the data identity guard. v0.1 Done
M2 Plans, approvals, projects and environments Plans with risk and approvals, durable operations, projects and environments with policies, the first CLI commands. v0.1 In development
M3 PostgreSQL high availability Patroni-based PostgreSQL clusters with endpoints, users and databases, certificate rotation and the create wizard. The v0.1 preview. v0.1 Planned
M4 Backups and restore Repositories, encrypted backups, point-in-time recovery, restores to a new cluster and restore drills. v0.2 Planned
M5 Monitoring, alerts and DBA tools Built-in metrics and alert rules, notifications, the SQL editor, schema browser and query insights. v0.3 Planned
M6 MCP and AI operations The MCP server, assistant setup, autonomy levels, limits per environment and kill switches. v0.3 Planned
M7 MongoDB, Valkey and Dragonfly MongoDB replica sets, Valkey with Sentinel, and Dragonfly (high availability in beta). v0.4 Planned
M8 MariaDB Galera Galera clusters with a preferred-writer endpoint and binlog point-in-time recovery. v0.5 Planned
M9 Garage S3 object storage Replicated Garage clusters, buckets, access keys and mirror backups. v0.6 Planned
M10 PostgreSQL day-2 completeness Major upgrades, disaster-recovery standby clusters, dedicated proxy servers, maintenance jobs, the advisor and data import. v0.7 Planned
M11 Platform and fleet operations The one-line installer, single sign-on, passkeys, teams, offline bundles, OS patching and webhooks. Release candidate. v0.8 Planned
M12 Hardening and general availability Security review, long-running chaos tests, upgrade paths and complete documentation. v1.0. v1.0 Planned

After v1.0

  • v1.1: Valkey cluster mode, MongoDB sharding, OAuth for MCP clients, online migration, blue/green upgrades, SCIM and ProxySQL.
  • v1.2: creating servers at cloud providers, starting with Hetzner, then AWS and GCP.
  • v1.3: a Kubernetes runtime and a Terraform and OpenTofu provider.
  • Later: SeaweedFS, ClickHouse and FerretDB as engine candidates.

How to read this roadmap

The order is deliberate: PostgreSQL goes first because it exercises every part of the system, and later engines reuse what it proves. If time gets short, scope is cut from a fixed list rather than shipping something half-tested: it starts with Dragonfly high availability and, near the end, moves whole engines to v1.1, MariaDB before MongoDB. Restore drills, PostgreSQL point-in-time recovery, the backup safety rules, human approval of AI changes, the audit log and failure testing are never cut.

Plans can change, and a roadmap is not a promise of delivery. When something moves, this page and the status marks across the site change with it.

Until releases exist, the interactive demo shows the console as it is planned, with simulated data.

See where it’s heading

The interactive demo runs the whole console on simulated data. Trigger a failover, approve an AI assistant’s plan, restore a backup to a point in time. Nothing to install.