Now in public beta · deltex.dev

SQL that runs
at the edge.

Deltex is a full PostgreSQL-flavored query engine compiled to WebAssembly, deployed across 70+ global edge nodes. No origin server. No cold starts. Your queries execute inside the CDN.

deltex · edge-node-fra-01 · wasm-v2
$ deltex query --region fra
30200µs
Typical query CPU time
70+
Global PoPs
140+
Scalar SQL functions
24ns
Per-row SIMD bitmap scan

The database was never
the right place for queries.

Traditional databases live in one region. Your users don't. Deltex moves the compute to them.

Zero cold-start latency

Wasm modules are pre-compiled and cached at every edge node. The first byte of your query hits a running engine, not a container waiting to boot. No warm-up. Ever.

🌐

Globally consistent, locally fast

Data is replicated across our globally distributed edge backbone. Reads resolve from the nearest PoP. Writes propagate in < 50 ms worldwide.

🗄️

Real SQL — not a subset

CTEs, correlated subqueries, window functions, lateral joins, JSON path operators, aggregates, recursive queries. If Postgres can do it, Deltex can do it at the edge.

🔒

No origin exposure

Your backend never touches the public internet. All query execution stays inside the edge network. No VPC, no firewall rules, no origin to DDoS.

How it works

Every query is a Wasm function. The CDN is the database runtime.

Client / Application
HTTP · HTTPS · WebSocket
Edge Compute Runtime
Wasm Runtime · SQL Parser · Planner · Executor
KV Store — globally replicated
IAD · ORD · LAXNorth America
LHR · FRA · AMSEurope
NRT · SIN · SYDAsia-Pacific
GRU · BOG · JNBEmerging

Each PoP runs an identical, stateless Wasm executor. KV Store provides durable, distributed state.

Full SQL.
Inside the CDN.

No feature flags. No SQL subset. The full engine — parser, planner, executor — ships inside the Wasm binary.

🧮

SIMD-Accelerated Columnar Engine

Wasm SIMD128 intrinsics power bitmap index scans at 2–4 ns/row. Columnar storage layout maximizes cache efficiency for analytical workloads.

wasm-simd128
📐

PostgreSQL-Flavored SQL

Full support for CTEs, recursive queries, window functions (RANK, LAG, LEAD, PARTITION BY), correlated subqueries, lateral joins, and FILTER clauses.

pg-compatible
🔣

140+ Scalar Functions

String, math, date/time, crypto (MD5, SHA-256), JSON, array, regex, encoding, and window functions — all compiled directly into the Wasm module.

built-in
🗂️

JSON Operators

Full ->, >>, @>, #> support. Query nested JSON documents inline with no preprocessing step.

jsonb
🔌

HTTP API + Driver SDKs

Query over plain HTTPS. JSON responses, typed SDKs for TypeScript, Go, Rust, and Python. No special protocol, no connection pooling, no driver installation.

REST · SDK
🛡️

Row-Level Security

Declarative RLS policies enforced inside the Wasm executor before any data leaves storage. Zero trusted surface — even the runtime can't bypass policies.

RLS

Deploy a schema.
Query in milliseconds.

1

Install the CLI

One command installs the Deltex CLI and authenticates against your account at deltex.dev.

2

Push your schema

Define tables in a schema.sql file and push. Deltex compiles your schema into the Wasm module and deploys it globally.

3

Query from anywhere

Use the HTTP API or any SDK. Queries route to the nearest PoP automatically — no configuration needed.

4

Ship to production

There is no step four. You're already in production. Deltex has no staging/production distinction at the infrastructure level.

shell
# Install the Deltex CLI
curl -fsSL https://deltex.dev/install | sh

# Authenticate
deltex login

# Create a new project
deltex init my-app
sql — schema.sql
CREATE TABLE events (
  id        BIGSERIAL PRIMARY KEY,
  user_id   UUID    NOT NULL,
  name      TEXT    NOT NULL,
  props     JSONB,
  ts        TIMESTAMPTZ DEFAULT now()
);

CREATE INDEX ON events (user_id, ts DESC);
typescript
import { createClient } from '@deltex/sdk'

const db = createClient({
  project: 'my-app',
  token:   process.env.DELTEX_TOKEN,
})

const rows = await db.query(`
  SELECT user_id,
         COUNT(*)                   AS total,
         MAX(ts)                    AS last_seen,
         props->>'plan'            AS plan
  FROM   events
  WHERE  ts > now() - INTERVAL '7 days'
  GROUP  BY user_id, plan
  ORDER  BY total DESC
  LIMIT  20
`)

console.log(rows.meta.total_us) // e.g. 1240

The database that lives
where your users do.

Deltex runs at 70+ edge nodes worldwide. There's no "us-east-1". Every PoP is production.

Open beta. Pricing →