Blog
BigQuery Migration with Xebia Axis: Modernize Your Data Platform on Google Cloud

Migration is a process, not a single step. Xebia Axis runs it end to end, from your legacy platform to BigQuery.
Migrating a data platform involves more than moving data. Existing workloads, complex dependencies and years of custom development make it hard to see the full scope, estimate the effort and keep the business running. No single tool covers that: a converter translates code, but it does not decide what moves first, prove the results still match, or handle what does not convert.
Xebia Axis runs the whole migration to BigQuery as one process, combining three things:
- A deterministic approach for everything that must be exact and repeatable: analysing the code, mapping dependencies, translating and applying tested rules.
- The power of agentic AI for the judgement calls: reviewing what automation drafted, completing what it could not, and repairing what fails.
- Deterministic quality gates between the steps: automated checks that decide when each step is done, never an AI agent.
Engineers stay in control of every decision that matters.
BigQuery Migration Process in Four Gated Steps

Each step ends at a quality gate; nothing reaches BigQuery without passing all four.
What Xebia Axis Brings to Your Migration
Four steps, each closed by its own quality gates:
- Assess the estate. Deterministic analysis reads the legacy code itself and finds the data assets, dependencies and workload complexity, with an effort and cost estimate, before any work begins.
- Plan in waves. Dependency mapping splits the estate into self-contained waves, each small enough to move and verify on its own. Engineers approve the plan before anything moves.
- Convert, then review with agentic AI. Deterministic translation and tested rules produce the first draft for BigQuery. AI agents review it and complete what automation could not. Each object must build on BigQuery before it moves on; when a gate fails, an agent repairs it from the gate's own error.
- Prove every wave. The original legacy processes run on generated test data to give the reference results. Every migrated table is compared with them, first locally and then on BigQuery. What still differs goes to an engineer, never silently through.
From Assessment to Migration
Xebia Axis supports two connected stages of the same process.
Xebia Axis for Readiness | 2 – 3 weeks
Assess existing assets, dependencies and complexity. Produce an inventory, findings, effort estimates and a wave plan to inform migration planning.
Xebia Axis for Migration | wave by wave
Convert legacy data models and ETL processes to BigQuery, one wave at a time. Each wave closes only when its quality gates pass; what fails is repaired or goes to an engineer for a decision.
Made for Google Cloud
Xebia Axis is built for BigQuery. Instead of reinventing them, it uses Google's own services where they fit: the BigQuery Migration Service for assessment and translation, the Data Validation Tool for the row-level comparison, and Gemini for AI-assisted translation and agents.

Xebia Axis carries the process and its gates; BigQuery and Google's services do the platform work.
Deterministic Where it Must be Exact, Agentic Where Judgement Helps, Gated at Every Step
Deterministic processing does the exact work: analysis, dependency mapping, translation and tested rules. AI agents handle what needs judgement: reviewing drafts, completing what automation could not, and repairing failures within a fixed budget.
The quality gates are code. They check that every migrated workload builds on BigQuery and produces the same results as the legacy platform, so an AI agent never decides whether the results match. Every stage produces reviewable outputs, and every decision that matters stays with the people responsible for the migration.
Customer Story: Migrating Oracle Database to Google Cloud
Step one of the process: mapping a complex Oracle data estate for a global organization in power technologies and energy systems.
The organization needed to understand a fragmented Oracle environment with extensive business logic and limited visibility. Xebia Axis ran the assessment: it mapped the estate, identified dependencies, scored complexity and fed a phased, costed migration roadmap. Migration is the next step of the same process.

The outcome: a mapped data estate, documented dependencies and an evidence-based, costed migration roadmap. The organization can now plan its migration from a clear view of the work, not from assumptions.
Why Xebia Axis for Data Warehouse Migration?
- End to end, with engineers in charge. One process from the first assessment to the final proof on BigQuery; AI agents work under supervision and people take the crucial decisions.
- No reinvented wheels. Xebia Axis orchestrates Google's own migration and validation services where they fit, and fills the gaps between them.
- No lock-in. Sources plug in the same way, and the AI agents can come from more than one provider.
- Cost-conscious. Most of the work runs locally at no cloud cost; BigQuery bills only the final check, under a fixed spending cap.
- Safe by design. Agents work in isolated test environments, never on production systems.
- Business knowledge kept. Rules and decisions are captured with their source, so they outlive the migration.
Go deeper in the Live Webinar: From Legacy System to Any Cloud Data Platform
This article describes the process. The webinar shows it applied to legacy migrations, step by step, including the complex, undocumented logic that usually blocks these projects. Xebia's data architects Marek Wiewiórka and Radosław Szmit take the four steps above further:
- Assess: what a full scan of your legacy estate tells you about workloads, lineage and cost.
- Plan: which work still needs people, and what a realistic timeline and budget look like before you approve a migration.
- Convert: how AI agents convert and validate SQL, ETL and orchestration logic in parallel, and what keeps it reliable.
- Prove: how to prove at scale that the new platform returns the same data, in the numbers and in their business meaning.
Free, live on 25 November 2026 at 14:00 CET, one hour online. The recording is available for registered participants.

Planning a move to Google Cloud?
Start with an assessment, the first step of the process, and see how Xebia Axis can carry your migration to BigQuery.
Talk to Xebia about your migration: xebia.ai/axis
Frequently Asked Questions about BigQuery Migration
What does a BigQuery migration involve beyond moving data?
Existing workloads, complex dependencies and years of custom development make the full scope hard to see. A converter translates code, but it does not decide what moves first, prove the results still match, or handle what does not convert. Xebia Axis runs the whole migration as one process.
How does Xebia Axis migrate a data warehouse to BigQuery?
In four steps, each closed by quality gates: assess the estate, plan in waves, convert and review with agentic AI, and prove every wave on BigQuery.
How long does the assessment take?
Xebia Axis for Readiness takes 2–3 weeks. It produces an inventory, findings, effort estimates and a wave plan that you can use for migration planning.
What is the role of AI in the migration?
AI agents handle the judgement calls: reviewing what automation drafted, completing what it could not, and repairing what fails. Deterministic processing does the exact work, and engineers approve the plan and take every decision that matters.
How do you prove the migrated data matches the legacy platform?
The original legacy processes run on generated test data to give reference results. Every migrated table is compared with them, first locally and then on BigQuery. The quality gates are code, so an AI agent never decides whether results match. What still differs goes to an engineer.
Which Google Cloud services does Xebia Axis use?
The BigQuery Migration Service for assessment and translation, the Data Validation Tool for row-level comparison, and Gemini for AI-assisted translation and agents.
Which legacy platforms can Xebia Axis migrate to BigQuery?
Xebia Axis is not tied to one source platform. Sources plug in the same way, so any legacy data warehouse or data platform, for example Oracle, Teradata, Snowflake or Amazon Redshift, goes through the same process: assess, plan, convert and prove.
Written by
Our Ideas
Explore More Blogs

AI Agent Studio: Swiftly Go From Data Contract to Certified Marketplace
Mayank Verma, Niels Zeilemaker
Contact













