Skip to content
Icelake Docs Icelake Docs Icelake Docs

Quickstart

Icelake is a fully managed, sovereign European data research workbench on an ultra-fast ClickHouse engine. This guide takes you from zero to answers in two stages — no servers on your side to run, nothing to install:

  1. In 60 seconds: load a ready-made sample dataset into a Sheet and ask it questions in plain language.
  2. In 5 minutes: stream live data from your own systems over the standard Prometheus, Loki, and OpenTelemetry endpoints and explore it in Grafana.

1. Get your Icelake account

Section titled “1. Get your Icelake account”

Sign up at app.icelake.eu or contact us if your team needs a pilot environment. Accounts are tenant-scoped and include everything — ingest endpoints, admin UI, pgwire SQL, AI MasterMind, dashboards — out of the box.

2. Your first data in 60 seconds — a Sheet

Section titled “2. Your first data in 60 seconds — a Sheet”

No API key and no tools needed — everything happens in the browser.

Download a sample (synthetic data — all companies, people and measurements are fictional):

SampleDownloadWhat’s inside
Business intelligencenorthpeak-outdoor-sales-2024-2025.zip (2.7 MB)122,000 orders of a fictional outdoor-gear brand — regions, channels, products, revenue, margin, returns
Scienceeu-air-quality-hourly-2025.zip (2.3 MB)140,000 hourly readings from 16 European stations — PM2.5, PM10, NO₂, O₃, temperature, wind
Medicalecg-12lead-hl7-obx.zip (2.9 MB)240,000 samples of four 12-lead ECGs at 500 Hz, flattened from HL7 v2 — one OBX segment per row

Then:

  1. Unzip the file — it contains the CSV and a short README.
  2. In the admin UI open Insights → Sheets → New Sheet, then click Load Data.
  3. Drop the CSV. Icelake uploads it straight to storage and detects delimiter, header and column types — review them and click Import. The sheet opens with your data a second later.
  4. Type a question into Ask your Sheet anything — for example:
    • Sales: “Show the monthly revenue per product category for 2025 as a chart. Which category peaks in winter?”
    • Air quality: “Which cities exceeded 15 µg/m³ PM2.5 on the most days?”
    • ECG: “Plot lead II of patient PAT-0002 for the first 4 seconds. Which finding is recorded for this patient?”

A Sheet with 122,000 sales orders and the AI chat charting monthly revenue per category

Every sheet is also a SQL table. Copy its name from the sheet header (SQL: sheet_…) and open Insights → SQL Workbench:

SELECT `Region`, toYear(`Order Date`) AS year, round(sum(`Revenue EUR`)) AS revenue
FROM sheet_your_sheet_id
GROUP BY `Region`, year
ORDER BY `Region`, year

More in the Sheets guide. Ready for live data? Continue below.

3. Create an API key

Section titled “3. Create an API key”

In the admin UI, open Data In → API Keys → Create key. You’ll get a client ID and a secret starting with ilk_. Copy the secret now — it’s shown exactly once. The same key authenticates every ingestion and query path.

Export them so the rest of this tutorial picks them up:

shell
export ICELAKE_CLIENT_ID="your-client-id"
export ICELAKE_API_KEY="ilk_..."

4. Send your first live signal

Section titled “4. Send your first live signal”

Any Prometheus agent, OpenTelemetry Collector, Fluent Bit, or curl can talk to Icelake. The shortest possible smoke test is one curl that pushes a log line into the Loki-compatible endpoint:

Push a log to Icelake
curl -X POST https://api.icelake.eu/loki/api/v1/push \
-u "$ICELAKE_CLIENT_ID:$ICELAKE_API_KEY" \
-H "Content-Type: application/json" \
--data-binary @- <<EOF
{"streams":[{"stream":{"app":"quickstart"},"values":[["$(($(date +%s) * 1000000000))","Hello from Icelake Quickstart"]]}]}
EOF

A 204 No Content means it landed. For real data flows, point your existing tools at the managed endpoints:

SourceEndpointDoc
Prometheus remote writehttps://api.icelake.eu/api/v1/prom/pushPrometheus
Loki push APIhttps://api.icelake.eu/loki/api/v1/pushLoki & LogQL
OpenTelemetry OTLP HTTPhttps://api.icelake.eu/otlp/v1/{logs,metrics}OpenTelemetry
MQTT (TTN or your broker)configured via the admin UIMQTT

All of them authenticate with the same client ID + ilk_… secret, either as HTTP Basic auth or as the pgwire username/password pair.

5. Launch Grafana pointed at Icelake

Section titled “5. Launch Grafana pointed at Icelake”

Already have a Grafana (cloud or internal)? Skip to the datasource config below. Otherwise, this docker-compose.yml plus a provisioning file brings up a local Grafana with Icelake pre-configured as both a Prometheus and a Loki datasource:

docker-compose.yml
services:
grafana:
  image: grafana/grafana:latest
  ports:
    - "3001:3000"
  environment:
    - ICELAKE_CLIENT_ID=${ICELAKE_CLIENT_ID}
    - ICELAKE_API_KEY=${ICELAKE_API_KEY}
  volumes:
    - ./grafana/provisioning:/etc/grafana/provisioning
grafana/provisioning/datasources/icelake.yaml
apiVersion: 1


datasources:
- name: Icelake Metrics
  type: prometheus
  access: proxy
  url: https://api.icelake.eu/prom
  basicAuth: true
  basicAuthUser: ${ICELAKE_CLIENT_ID}
  secureJsonData:
    basicAuthPassword: ${ICELAKE_API_KEY}
  isDefault: true


- name: Icelake Logs
  type: loki
  access: proxy
  url: https://api.icelake.eu/loki
  basicAuth: true
  basicAuthUser: ${ICELAKE_CLIENT_ID}
  secureJsonData:
    basicAuthPassword: ${ICELAKE_API_KEY}

Run docker compose up -d. Grafana picks up the datasources on startup — no click-through config needed.

6. Explore in Grafana Metrics Drilldown

Section titled “6. Explore in Grafana Metrics Drilldown”

Grafana Metrics Drilldown exploring Home Assistant metrics stored in Icelake

Open http://localhost:3001, sign in (default admin / admin), and head to Explore → Metrics. Grafana reads labels and series directly from Icelake’s Prometheus-compatible /prom/api/v1/* endpoints, so Metrics Drilldown, ad-hoc PromQL, and saved dashboards all work out of the box. Switch the datasource to Icelake Logs for LogQL against your ingested logs.

Next steps

Section titled “Next steps”