BigQuery used to be where reports lived. Someone ran a query once a month, glanced at a dashboard, and moved on.
That’s not what it is anymore.
Modern AI has turned BigQuery from a place you analyze data into a place where AI runs on it. Teams call Gemini straight from SQL to pull structured fields out of contracts and support tickets. They store vector embeddings next to their tables, so a chatbot or search box answers from live company data instead of a separate vector database. They point BigQuery at PDFs, receipts, and images in Cloud Storage and query them like rows. And when autonomous agents act for the business, such as deciding whether to offer a churned customer a discount, they check BigQuery first for the customer’s spend history and recent activity.
It’s also still where the financial records sit, where BI pipelines run, and where the training data for your ML models comes from. Every one of those AI workloads depends on it being correct and available.
Lose it, and the dashboards go dark. The models go stale. The agents start guessing, and the audits fail. Your business can no longer make informed decisions.
“No problem,” you think. “Google Time Travel solves this.” Except it doesn't do what you're hoping it does. Google Time travel is useful for short-term accidents. It is not a backup strategy.
Google Time Travel rolls a table back seven days, inside the same account that holds the original data. That covers a dropped table or a bad query. It leaves a compromised account, a ransomware event, or an audit that requires records from three years ago with no independent copy to recover from. Rubrik Google BigQuery Data Protection closes that gap with automated, SLA-driven backups of your datasets, tables, views, and routines, stored immutably outside your production account and restored as a working, queryable dataset.
Time Travel Isn't A Backup Strategy
Google Time Travel lets you roll a table back to an earlier state. It works well for what it's built for: a dropped table, a bad query, a quick accidental delete. But there’s a problem. Time Travel caps out at 7 days. After that, there's nothing left to roll back to.
It's also mutable. Time Travel data lives inside the same project as the table it's protecting, and any admin with the right IAM role can affect it. If the underlying data disappears, whether that's a bad actor, a ransomware event, or a storage cleanup that went further than intended, there's no independent copy sitting somewhere else to fall back on.
Google gives you table snapshots too, but those are manual—no scheduling, no automation, no lifecycle policy. Someone has to build and maintain these snapshots. Google is upfront about this: infrastructure is their responsibility, your data is yours. That's the shared responsibility model working exactly as designed.
Rebuilding Is Not the Same As Recovering
Say the data itself is still out there somewhere and you're facing a rebuild instead of a clean recovery. A BigQuery dataset isn't one table. It's hundreds or thousands of tables, partitioned tables, materialized views, routines, each with its own schema and dependencies.
Reconstructing that by hand takes real engineering time. It generates compute and egress costs along the way, and some transformed data can't be rebuilt at all because the pipeline that produced it doesn't run in reverse. While that rebuild is happening, every BI dashboard, every ML pipeline, and every compliance report built on those tables is offline for as long as it takes.
The Solution: Rubrik Google BigQuery Data Protection
Rubrik Google BigQuery Data Protection extends Rubrik's Google Cloud data protection platform to BigQuery as a fully managed workload on Rubrik Security Cloud. Rubrik automatically discovers your datasets and tables, applies SLA-driven backup policies, and archives data immutably to Google Cloud Storage or Rubrik Cloud Vault, outside your production account and/or region.
This isn't just the rows in a table. Rubrik protects views, materialized views, routines, and row-level security policies, and recovers them together as part of the dataset. When something goes wrong, what comes back is queryable and usable immediately, not a pile of files waiting on someone to rebuild the metadata by hand.
How It Works
Protecting a BigQuery dataset means capturing more than the data sitting in tables. Rubrik also has to track schema, partitioning and clustering configuration, views, routines, and access policies, and know how to put all of it back together in a way BigQuery recognizes on the other end.
A few things about how that works:
Rubrik protects at the dataset level, but SLA assignment can be at the Project, Label, or Dataset hierarchy, and coverage extends automatically to new tables and datasets when assigning SLAs to entire projects or based on label rules.
Incremental backups use BigQuery's own CHANGES() function to capture row-level inserts, updates, and deletes, rather than exporting entire tables on every run.
Recovery works at the dataset level or the table level, so you can bring back exactly what broke instead of an entire environment.
Backups land outside your Google Cloud account, whether that's your own Google Cloud Storage or Rubrik Cloud Vault, out of reach of anything that can touch production. Ensure your backups stay within a different region or different project
Since Rubrik can archive to Google Cloud Storage's cold storage tiers, long-term retention doesn't mean paying BigQuery storage rates to keep data you rarely query.
One Platform For Your Full Google Cloud Estate
BigQuery doesn't run in isolation. It sits inside the same Google Cloud projects and the same IAM boundaries as the rest of your environment, fed by pipelines whose code usually lives in GitHub, and by services you're likely already running elsewhere.
BigQuery now sits alongside GCE, Cloud SQL, and the GitHub repositories that build your pipelines, all under one platform—with the same SLA policies, RBAC, and compliance reporting. Instead of a separate tool with its own login and its own report to check, it's one more workload under a policy framework you already have running.
Try It Yourself
Google Time Travel and table snapshots solve real, short-term problems. They were never built to survive a compromised account or a compliance audit asking for data from three years ago. Rubrik Google BigQuery Data Protection closes that gap: automated, SLA-driven backup and recovery for your tables, views, routines, and schema, stored somewhere the original failure can't reach, and restorable to a working, queryable dataset without anyone touching the catalog by hand.
Ready to give it a shot? Take our BigQuery protection for a spin by taking one of our self-guided, hands-on labs on Rubrik Explore.