Products · Databases

Managed cloud databases
with a real client built in.

Provisioning a database is the easy part. What you actually spend your day in is the tooling — so Darwa ships the browser, the query editor, the index manager, and the assistant with every instance. No DBeaver, pgAdmin, TablePlus, or SSH tunnel required.

Create a database See pricing

— No client to install— Basic + Premium zones— Backups included

storefront · postgres 17healthy

▾ storefront
  ▾ Tables            24
      orders          1.2M
      customers       84k
      line_items      4.1M
      payments        1.1M
  ▸ Views             6
  ▸ Materialized      2
  ▸ Indexes           41
  ▸ Functions         9
  ▸ Triggers          4
  ▸ Sequences         24
  ▸ Extensions        3
  ▸ Users             5

query-1.sqlslow-orders.sqlsaved · monthly-revenue

  1  SELECT c.country, count(*) AS orders, sum(o.total) AS revenue
  2  FROM orders o
  3  JOIN customers c ON c.id = o.customer_id
  4  WHERE o.created_at >= now() - interval '30 days'
  5  GROUP BY c.country
  6  ORDER BY revenue DESC LIMIT 5;

Run ⌘↵explainformatsavehistory

	country	orders	revenue
1	DE	18,402	912,441.20
2	US	16,118	884,205.65
3	IN	14,904	402,118.40
4	AE	9,221	388,904.00
5	SG	7,860	241,660.75

rows 5time 41 msplan index scanexport CSV · JSONconnection pooled

deploy engines workspace explorer query operate scale secure assistant connect pricing

darwa ▸ database ▸ create

One click, then a connection string.

Pick an engine, a version, and a region. The instance, credentials, TLS, pooling, and the first backup schedule are configured before the page finishes loading.

Choose the versionPin a major version and stay there, or opt into automatic minor upgrades in a maintenance window you set.

Choose the regionPut the database in the same region as the service that reads it, on the private network.

Scale without re-platformingStart at 256 MB and upgrade through 256 GB as traffic, connections, and working data grow.

Instant connection stringsInjected into your services as environment values — never pasted into a repository.

Connection poolingPgBouncer-style pooling in front of PostgreSQL and MySQL, so a scaled-out worker pool cannot exhaust connections.

Import and exportRestore a dump on creation, or export to SQL, CSV, or JSON at any time.

Created in 34 seconds

storefront · postgres 17ready in 34s

Regioneu-centralsame region as app-backendset

InstanceGrowth · 4 GB · 1 vCPUmanaged compute with predictable resourcesset

TLSsslmode=require enforcedcertificate managed and rotated

PoolingPooler on :6543 · 200 clientsworkers cannot exhaust connections

BackupsDaily 02:00 · PITR 14 daysrestore to any second in the window

Public accessClosedprivate network only — open it deliberately

DATABASE_URL injected into 2 servicesfirst backup done

darwa ▸ database ▸ engines

Five database options, each with the full experience.

PostgreSQL, MySQL, MariaDB, document storage via FerretDB on PostgreSQL, and Redis / Valkey are available now. PostgreSQL JSONB also supports document-first workloads without a separate engine.

Database	Type	Status
PostgreSQL	Relational (SQL) · JSONB documents	Available
MySQL	Relational (SQL)	Available
MariaDB	Relational (SQL)	Available
Document (FerretDB)	MongoDB drivers on PostgreSQL	Available
Redis / Valkey	In-memory key-value	Available
SQLite	Embedded SQL	Coming soon
ClickHouse	Analytics (OLAP)	Coming soon
TimescaleDB	Time-series SQL	Coming soon
Neo4j	Graph	Coming soon
Cassandra	Wide-column	Coming soon
ScyllaDB	Wide-column	Coming soon
Couchbase	Document + key-value	Coming soon
Apache Doris	Real-time analytics	Coming soon
Elasticsearch / OpenSearch	Search & indexing	Coming soon

darwa ▸ database ▸ workspace

One console, shaped to each database.

Every database gets the same navigator, tabbed object workspace, editor, AI, recovery, security, and connection surfaces. The nouns and tools adapt to the engine instead of forcing document and key-value data into SQL screens.

PostgreSQL

SQL + JSONB

Schemas · tables · views · functions · triggers · extensions

SQL editor, JSONB grid, explain plans, ER diagrams

MySQL

Relational SQL

Schemas · tables · views · routines · triggers · events

SQL editor, row grid, explain plans, ER diagrams

MariaDB

Relational SQL

Schemas · tables · views · routines · triggers · events

SQL editor, row grid, explain plans, ER diagrams

Document

FerretDB · MongoDB drivers

Collections · documents · indexes · aggregations · users

Document grid, JSON editor, filters, aggregation pipelines

Redis / Valkey

In-memory key-value

Keyspaces · keys · hashes · lists · sets · streams

Key browser, TTL and memory inspector, command editor

public / tablesorders1,204,882 rows · 684 MB

DataBColumns / StructureBConstraintsBForeign keysBIndexesBTriggersBDDLBAIBDependenciesAPermissionsAER miniAPartitionsA

id	customer_id	status	total	created_at
ord_1204882	cus_84102	paid	$128.40	2026-08-28 12:41
ord_1204881	cus_29214	pending	$74.00	2026-08-28 12:39
ord_1204880	cus_61008	paid	$241.65	2026-08-28 12:37

BNavigate + edit

Full object tree, search, row counts and sizes; inline data editing with guarded writes and explicit production confirmation.

BQuery + understand

Multi-tab native editor, schema autocomplete, history, saved queries, explain plans, results charts, CSV and JSON export.

BAI with context

Generate, optimize, explain, and find indexes against the real schema. DDL is proposed for review and never auto-applied.

BMove data

Import CSV, JSON, or SQL; export data and schemas. Engine-aware document grids expand nested JSON without flattening it.

BOperate safely

Backups, point-in-time recovery, slow-query insights, metrics, alerts, audit events, users, roles, scoped keys, and secrets.

BWork everywhere

Command palette, global object search, multi-tab workspace, split view, keyboard shortcuts, and light or dark appearance.

ADesign visually

Visual query builder, interactive ER neighbors, chart suggestions, and saved query dashboards.

ACompare + deploy

Schema and data comparison, database-to-database transfer, reviewed deploy requests, and mock data generation.

AOperate deeply

Active sessions, permissions and RLS policies, partitions, maintenance deferral, replicas, and failover topology.

darwa ▸ database ▸ explorer

Every object in the database, in the browser.

Not a metrics page with a connection string at the bottom. The full object tree, editable, with row counts and sizes where they matter.

SQL engines

▾ Database
    Tables            browse · edit · truncate
    Views             definition · dependencies
    Materialized      refresh · last refreshed
    Indexes           size · usage · unused
    Functions         source · arguments
    Triggers          table · event · timing
    Sequences         current value · owner
    Extensions        install · version
    Users             roles · grants

Document · FerretDB

▾ Database
    Collections       documents · size · shape
    Documents         view · edit · validate
    Indexes           keys · usage · TTL
    Aggregations      pipeline builder · explain
    Users             roles · scoped access

FerretDB speaks the MongoDB wire protocol on PostgreSQL, so existing MongoDB drivers connect unchanged. Documents open in a JSON editor with schema shape inferred from a sample, so inconsistent fields are visible rather than surprising.

Schema browser

Columns, types, defaults, nullability, foreign keys, and what references this table.

Index manager

Create, drop, and rebuild — with size, scan counts, and a plain list of indexes nothing has used.

Users and grants

Create a read-only role for a reporting tool without opening a psql session.

darwa ▸ database ▸ query editor

A query editor you would actually use.

Tabs, autocomplete on your real schema, history, saved queries, and exports. The editor at the top of this page is the product, not an illustration.

Autocomplete on your schemaTable and column names, joins suggested from foreign keys, functions with their signatures.

Multiple tabsKeep a migration draft, a debugging query, and a report open at once.

Query historyEvery statement you ran, with duration and row count, searchable — including the one you forgot to save.

Saved queriesName a query, share it with the team, run it against staging or production.

Export resultsCSV or JSON, straight from the result grid.

Guarded writesStatements without a WHERE clause ask for confirmation, and production requires an explicit toggle.

Execution plan, in words

Seq scan on line_items1,740 ms

4.1M rows read to return 128 · no usable index

Hash join · customers124 ms

Spilled to disk — work_mem too low for this join

Index scan on orders18 ms

orders_created_at_idx · already optimal

Proposed fix

CREATE INDEX CONCURRENTLY ON line_items(order_id);

p95 1.9 s → 96 msindex size 84 MBwrite cost +3%

Run nowSchedule for 02:00Show raw plan

The plan is summarised in language you can act on, with the raw plan one click away for when you want it.

darwa ▸ database ▸ operations

Backups you have actually tested.

Daily snapshots plus point-in-time recovery to any second in the retention window — and a restore that lands in a new instance, so you can verify it before you touch production.

Automatic backups

Daily, encrypted, in a window you choose. Retention runs from 7 to 35 days depending on the plan.

Point-in-time recovery

Rewind to any second within retention — the state just before a bad migration, not just last night.

Restore to a new instance

Recovery creates a separate database by default. Check it, then repoint your services.

Automatic minor upgrades

Patch versions applied in your maintenance window, with the changelog and a one-click deferral.

Health checks and alerts

Connection saturation, replication lag, disk pressure, and long transactions — alerted before they page you.

Storage autogrowth

Disk expands before it fills, and you are told what it cost rather than finding out on the invoice.

Point-in-time recovery · 2 Aug 2026

base02:00 snapshotwal12h 12m replayedtarget14:12:09now14:41 live

Restored tostorefront-restore-0802a new instance — production untouchedsafe

Verified1,204,882 orders · last row 14:12:08row counts checked against the target timematch

Divergence41 rows written after this pointexport them, or accept the loss knowinglyreview

Promote to primaryExport the 41 rowsDiscard restore

darwa ▸ database ▸ scaling

Grow the instance, add readers, survive a failure.

Vertical scaling on a rolling restart, read replicas where the engine supports them, and automatic failover with the connection string unchanged.

Vertical scalingMore CPU, memory, or disk applied in a maintenance window — typically under 60 seconds of failover.

Read replicasPostgreSQL, MySQL, and MariaDB. Point reporting and search at a replica, keep writes on the primary.

High availabilityA synchronous standby in another availability zone, promoted automatically when the primary fails.

Automatic failoverYour services keep the same host name; the pooler routes to the new primary.

Replication lag visibleLag per replica in seconds and bytes, alerted when a replica falls behind your threshold.

Topology · storefront

Primaryeu-central · 8 GB
41% CPU · accepts writes

Standbyeu-central-b · synchronous
lag 0 ms · auto-promote

Replica · reportingeu-west · async
lag 240 ms

Replica · read APIap-south · async
lag 1.1 s

failover automaticlast tested 12 Jul · 38 shost name unchanged on promote

darwa ▸ database ▸ security

Private by default, public only if you insist.

A new database has no public endpoint. It is reachable over the private network by your own services, and nothing else.

Control	What it does	Default
Private networking	Reachable only from your services in the same region	On
TLS connections	sslmode=require enforced, certificates managed and rotated	On
IP allowlists	If you do open a public endpoint, only listed addresses can reach it	Closed
Role-based access	Per-role grants, with read-only roles created in two clicks	On
Secrets management	Credentials injected at runtime, rotated without a code change	On
Audit trail	Who connected, who ran DDL, who exported data, and when	On

darwa ▸ database ▸ assistant

AI where the database work happens.

Ask from a table, a query, or the whole database. The assistant uses the permitted schema, indexes, query plans, and health signals so every answer names the real object and proposes a reviewable next step.

Object workspace

Ask about this table

Explain columns and relationships, generate useful queries, detect risky types, and design a safe table migration without losing the selected object.

AI assistant · coming soon

SQL editor

Generate and rewrite SQL

Turn a question into engine-aware SQL, explain an existing statement, fix errors, and optimize it against the live schema.

AI assistant · coming soon

Execution plan

Explain the expensive step

Name the scan, join, spill, or lock dominating runtime and compare the measured plan before and after a proposed change.

AI assistant · coming soon

Index advisor

Rank missing indexes

Prioritize indexes by recoverable query time while showing disk growth, write overhead, lock behavior, and unused-index candidates.

AI assistant · coming soon

Database health

Diagnose why it is slow

Correlate connections, locks, long transactions, vacuum state, storage pressure, replication lag, and recent schema changes.

AI assistant · coming soon

Safety and governance

Propose first, apply explicitly

Show SQL or DDL before execution, identify affected objects, respect role permissions, redact secrets, and record approved changes in the audit log.

AI assistant · coming soon

“why is my database slow?”3 causes

line_items(order_id) has no index62% of total time

4.1M row scans across 1,204 queries in the last hour

Connections 186 of 20040 ms queued

The pooler is holding requests before they reach the database

Autovacuum behind on orders18% dead tuples

Planner estimates drift as dead rows accumulate

Fix for the first cause

CREATE INDEX CONCURRENTLY ON line_items(order_id);

p95 1.9 s → 96 msdisk +84 MBlocks none — concurrent build

Run nowSchedule for the maintenance window

“Generate SQL for this table.”Describe what you want in a sentence and get a statement written against your actual columns.

“Optimize this query.”A rewritten version, the reason it is faster, and the measured difference on your data.

“Find missing indexes.”Ranked by time recoverable, with the write cost and disk cost of each one stated.

“Explain this execution plan.”The plan in plain language, with the specific step that dominates the runtime.

“Design a safe migration.”Generate reversible schema steps, flag locking operations, and show the affected tables before anything runs.

“Check data quality.”Find unexpected nulls, duplicates, orphaned relationships, and unusual value distributions with queries you can inspect.

“What changed?”Schema and index changes correlated with the deploy and the latency shift that followed.

Scoped contextUse metadata and telemetry allowed by the current role. Row values are included only when the user deliberately selects them.

Guarded by defaultThe assistant proposes DDL; you run it. Nothing is applied to a database without an explicit action.

darwa ▸ database ▸ connect

Every front door, in one Connect panel.

Copy a native connection once, or issue a scoped interface for an application, agent, automation, or teammate. Availability is shown per engine before a credential is created.

Direct

Native credentials and TLS connection string

All five

Pooled

Managed pooler endpoint for bursty applications

SQL + Document

REST

Scoped base URL and expiring API key

All five

GraphQL

Schema-aware endpoint for supported relational data

SQL engines

MCP

Scoped tools URL for agents and IDEs

All five

Automations

Recipes for n8n, Zapier, and Make

All five

Keys can be temporary or permanentSet expiry, scope, budget, and rotation independently. Secrets are shown once, encrypted at rest, and revocable without changing the database.

darwa ▸ database ▸ pricing

Basic and Premium database pricing, in one place.

Choose a zone to see its monthly database plans and exact storage and network rates. Premium zones use a separate zone rate.

Live catalog unavailable

Zone pricing is temporarily unavailable.

Please try again shortly. No estimated or stale prices are shown in its place.

Get started

Create a database and open the editor.

The free instance has no expiry date, keeps its backups, and comes with the same explorer, editor, and assistant as production.

Create a database Connect it to a service

Managed cloud databaseswith a real client built in.

One click, then a connection string.

Five database options, each with the full experience.

One console, shaped to each database.

PostgreSQL

MySQL

MariaDB

Document

Redis / Valkey

Every object in the database, in the browser.

A query editor you would actually use.

Backups you have actually tested.

Grow the instance, add readers, survive a failure.

Private by default, public only if you insist.

AI where the database work happens.

Every front door, in one Connect panel.

Basic and Premium database pricing, in one place.

Zone pricing is temporarily unavailable.

Create a database and open the editor.

Managed cloud databases
with a real client built in.