In product
BigQuery · Storage Intelligence

Logical vs physical.
Time travel, fail-safe, owners.

Anavsan is the Workload Governance Platform for Snowflake and BigQuery. Powered by APEX, it traces waste (not spend) to a workload and owner, assigns the fix, enforces a decision through your review process, and proves closure from the data.

Dataset-level logical billing (default, uncompressed; time travel and fail-safe included) versus physical (compressed; time travel and fail-safe billed separately at active physical rates). Active versus long-term — about 50% after 90 days unmodified. Signals from INFORMATION_SCHEMA.TABLE_STORAGE. Never table contents.

Book a Demo Start Free Trial
The problem

Storage is billed two ways.
Most teams only see one line.

Google’s invoice totals storage. It does not tell you which datasets are unowned, which tables have no expiration, or why physical billing plus a long time travel window became a surprise.

Logical default vs physical surprise

Logical bills uncompressed bytes and folds time travel and fail-safe into the rate. Physical bills compressed active storage; time travel and fail-safe are extra, at active physical rates.

Time travel 2–7 days; fail-safe is 7 and fixed

Time travel is configurable from 2 to 7 days. Fail-safe is 7 days and not user-configurable. Physical model plus a long window is where the bill jumps — not Snowflake-style clone sprawl.

Unowned datasets and missing expiration

Datasets without an owner or table expiration keep growing. If ownership is not knowable from metadata, the finding stays needs owner.

How APEX applies

Storage waste gets a dataset and an owner.

Same loop as the Snowflake Storage Intelligence sibling — not a clone. BigQuery objects only. INFORMATION_SCHEMA.TABLE_STORAGE. The Private Knowledge Graph remembers ownership — it does not AutoApply.

✓
Logical vs physical billing model made explicit
✓
Time travel window and fail-safe called out
✓
Active vs long-term (~50% after 90 days)
✓
Unowned datasets and missing expiration
✓
Decision through your review process
How APEX applies

Trace → Assign → Enforce → Prove

Four stages. Dataset and table storage objects. No AutoApply. No Snowflake dollar rates copied onto this page.

01 · Trace

Waste on datasets

Rank unowned datasets, missing expiration, and physical-model time travel plus fail-safe — not the largest storage line on the invoice.

02 · Assign

A named owner

Resolve the dataset from evidence. If ownership is not knowable, the finding stays needs owner. The Private Knowledge Graph is ownership memory — it does not AutoApply.

03 · Enforce

Via review / PR

Expiration, billing model, or time travel window changes go through your review process until applied, deferred with a reason, or dismissed.

04 · Prove

Metric recovered

The finding closes when billed storage — logical or physical, including time travel and fail-safe on physical — returns to the expected band. Closure is from the data.

Honest vs Google native

Google already shows X.
Anavsan closes Y.

We do not replace INFORMATION_SCHEMA.TABLE_STORAGE or the billing console. We attach storage waste to an owner and a closed decision.

Google already shows

Bytes, billing model, windows

  • INFORMATION_SCHEMA.TABLE_STORAGE — logical and physical bytes
  • Dataset storage billing model: logical (default) vs physical
  • Time travel window (2–7 days) and fail-safe (7 days, fixed)
  • Active vs long-term storage after 90 days unmodified (~50%)
Anavsan closes

Waste, owner, review, proof

  • Rank unowned datasets and missing expiration as waste, not spend
  • Flag physical model + long time travel as a surprise-bill pattern
  • Assign a named owner, or leave the finding as needs owner
  • Enforce through your review process — never AutoApply — and prove the storage metric recovered
FAQ

Storage questions

In product. Metadata only. BigQuery billing — not a Snowflake clone.

How does BigQuery bill storage — logical or physical?
Both, by dataset. Logical is the default: uncompressed bytes, with time travel and fail-safe included. Physical bills compressed active storage; time travel and fail-safe are billed separately at active physical rates. Active data unmodified for 90 days moves to long-term at about 50% of the active rate.
What are BigQuery time travel and fail-safe windows?
Time travel is 2–7 days. Fail-safe is 7 days and not user-configurable. On the physical model, a long time travel window plus fail-safe is a common surprise bill — those bytes are billed separately at active physical rates.
Does Anavsan read BigQuery table contents?
No — operational metadata such as INFORMATION_SCHEMA.TABLE_STORAGE only. Anavsan never reads, copies, or moves table contents.
Is this a clone of Snowflake Storage Intelligence?
No — sibling, not a clone. The Snowflake Storage Intelligence page covers Snowflake Time Travel, Fail-Safe, and clone sprawl in credit terms. This page uses BigQuery billing models only. Same APEX loop. The Private Knowledge Graph assigns ownership — it does not AutoApply.
How is storage intelligence different from dataset lineage?
Storage is billing. Lineage is jobs to tables to owners. This page covers logical vs physical, time travel, fail-safe, and unowned datasets. Dataset lineage uses JOBS referenced_tables — not Dataplex. Cost waste is on cost governance.

See storage waste ranked
on your project.

BigQuery Storage Intelligence is in product. Book a 30-minute demo or start a trial — Trace → Assign → Enforce → Prove, on datasets and billing models.

Book a Demo Start Free Trial
BigQuery hub → Cost governance → Dataset lineage → APEX engine → Snowflake storage sibling →