Analytics

A dedicated graph analytics engine per project — Cypher and SQL over your own isolated knowledge graph, with scale-to-zero pricing

Overview

Scalix Analytics gives every project its own dedicated graph analytics engine — not a shared cluster. The first query boots a hardware-isolated microVM for your project (cold start ~2.5s, warm queries in milliseconds); after 5 idle minutes it scales to zero and you stop paying for compute. Your graph lives on a persistent per-project volume, so the data is there when the engine wakes back up.

You can query the same data two ways:

• Cypher — full read-only graph queries (MATCH, traversals, aggregations)
• SQL — basic SELECT statements, translated to Cypher by the engine

Analytics is available on Pro plans and above, behind a per-service agreement you accept once in Console → Billing → Services (you can set a monthly spend cap there too, as an alert or a hard stop).

Quick start (60 seconds)

1. Open Console → your project → Analytics and enable the service if prompted.
2. Go to the Load Data tab and click Load sample — a tiny social graph loads through the real ingest API and the Graph Explorer opens.
3. Run your first query:

MATCH (a:Person)-[r:KNOWS]->(b:Person)
RETURN a.name, b.name, r.since

Querying via the API

Use a project-scoped API key with the prime:read scope:

curl -X POST https://api.scalix.world/v1/prime/query \
  -H "Authorization: Bearer $SCALIX_API_KEY" \
  -H "Content-Type: application/json" \
  -d '{"query": "MATCH (n) RETURN n LIMIT 10", "language": "cypher"}'

The response is a QueryResult:

{
  "columns": ["n"],
  "rows": [{"n": {"_id": "0:0", "_labels": ["Person"], "name": "Ada"}}],
  "row_count": 1,
  "execution_time_ms": 4.2
}

For SQL, set "language": "sql". Only basic SELECT is supported today — anything the translator can't handle returns an honest 400 SQL_NOT_SUPPORTED rather than wrong results. Writes are rejected with 403 READ_ONLY on the query surface; data enters only through ingest.

Loading data

CSV upload

POST /v1/prime/ingest with a raw CSV body (header row required) and the prime:write scope. Parameters travel as headers:

# nodes.csv:  id,name,city
curl -X POST https://api.scalix.world/v1/prime/ingest \
  -H "Authorization: Bearer $SCALIX_API_KEY" \
  -H "Content-Type: text/csv" \
  -H "x-scalix-ingest-kind: nodes" \
  -H "x-scalix-ingest-label: Customer" \
  --data-binary @nodes.csv
 
# edges.csv: first two columns are the from-id and to-id
curl -X POST https://api.scalix.world/v1/prime/ingest \
  -H "Authorization: Bearer $SCALIX_API_KEY" \
  -H "Content-Type: text/csv" \
  -H "x-scalix-ingest-kind: rels" \
  -H "x-scalix-ingest-rel-type: PURCHASED" \
  -H "x-scalix-ingest-from: Customer" \
  -H "x-scalix-ingest-to: Product" \
  --data-binary @edges.csv

Rules the loader applies:

• Every value loads as a string (cast in your queries as needed).
• For nodes, the id column — or the first column if there's no id — becomes the node key. Loading the same key twice fails the upload (INGEST_FAILED), it never silently overwrites.
• For relationships, the first two CSV columns are the from/to node keys; any further columns become relationship properties.
• Repeated uploads to the same label append.

Import from your database

One click in the Load Data tab (or POST /v1/prime/import-nova) snapshots your project's ScalixNova Postgres into the graph: every table becomes a node label, every foreign key becomes a relationship type ({table}_{column}_{referenced_table}). The API key needs both prime:write and a database read scope. Poll GET /v1/prime/import-nova/{job_id} for progress. It's a one-shot snapshot (not continuous sync); re-importing the same rows fails on duplicate keys by design.

Pricing

Three meters, all visible live in the console (rates come from the billing catalog — what you see there is what bills):

Illustrative: an evening of exploration costs a few cents; a team project querying all workday ≈ $15–40/month; an always-on integration lands near $300/month — at parity with managed graph databases that never scale to zero.

Limits (v1)

Error codes