Snowflake cost monitoring in the same ledger as your cloud

What CloudQuell ingests

Three views from SNOWFLAKE.ORGANIZATION_USAGE — the authoritative billing record, the one that matches your invoice.

Dollars, daily

USAGE_IN_CURRENCY_DAILY — spend by account, region, and service type, per day. This is the invoice-grade number.

Credits by warehouse

WAREHOUSE_METERING_HISTORY — compute and cloud-services credits per warehouse, the grain that makes warehouse-level cost attribution possible.

Effective credit rates

RATE_SHEET_DAILY — your negotiated per-credit rates, so credits reconcile to dollars instead of a list-price guess.

What it doesn't read

Only usage metadata — never table data, query text, or results. Snowflake object tags are not imported, so Snowflake sits outside the tag-coverage view by design.

What the data looks like normalized

Credits and the dollars they cost land on the same row, in the same ledger as your cloud and AI spend.

How it connects

A dedicated read-only service user, key-pair authenticated, running on its own extra-small warehouse.

Credential

Key-pair (RSA) authentication on a CLOUDQUELL service user and role. CloudQuell generates the key pair and stores the private key in AWS Secrets Manager — you never handle it.

Access scope

Read-only usage metadata only. Queries run on a dedicated CLOUDQUELL_WH extra-small warehouse that auto-suspends after 60 seconds, so the connector's own footprint is negligible.

Prerequisites

An ORGADMIN-enabled organization account and an ACCOUNTADMIN to run the one-time script that provisions the user, role, and grants.

Time to first data

About ten minutes end to end, then a 12-month backfill. Daily sync after that; dollar figures finalize up to ~72 hours behind to match Snowflake's own end-of-month adjustments.

What it unlocks

Credits and dollars, allocated and budgeted on the same model as the rest of your spend.

• Warehouse attribution

  Daily spend and credits per warehouse, a blended $/credit rate, and a warehouse trend — so the expensive warehouse has a name, not just a number.

• Cost allocation

  Route warehouse spend to cost centers with rules; compute and cloud-services credits split by daily share and marked estimated where they are apportioned.

  How allocation works →

• Budgets

  Set budgets on those cost centers and get warned before a runaway warehouse blows past them, in the channel your team already watches.

• Agent access

  Query Snowflake spend from Claude, Cursor, or any MCP client alongside your cloud and AI cost — the same ledger, one question.

Common questions

Does CloudQuell track Snowflake credits or dollars?

Both, together. It reads credits per warehouse from WAREHOUSE_METERING_HISTORY and dollars from USAGE_IN_CURRENCY_DAILY, and reconciles them through your effective rate sheet — so you see a warehouse's credits and what those credits actually cost, side by side, next to the rest of your cloud bill.

Can CloudQuell see my Snowflake data?

No. The service user is scoped to usage metadata in ORGANIZATION_USAGE only — never table data, query text, or query results. It runs on a dedicated extra-small warehouse that auto-suspends after a minute, and the private key lives in AWS Secrets Manager, not in CloudQuell's database.

Does it attribute cost per warehouse?

Yes — warehouse is the primary attribution grain. Allocation rules route each warehouse's spend to a cost center; where compute and cloud-services credits have to be apportioned, the split is by daily credit share and flagged as estimated so the assumption is visible.

Are Snowflake object tags supported?

Not today. Snowflake object tags created with CREATE TAG are not ingested, so Snowflake sits outside CloudQuell's tag-coverage view; warehouse-based allocation rules are the attribution mechanism instead. (The cloud tag-coverage metric applies to AWS.)