Workload

Oracle stored procedures & UDFs to BigQuery

Re-home PL/SQL business logic—procedures, functions, packages, and trigger-driven side effects—into BigQuery routines with an explicit behavior contract and test harness so retries and backfills don’t change outcomes.

Book assessment Explore workloads

Quick answer

Re-home PL/SQL business logic—procedures, functions, packages, and trigger-driven side effects—into BigQuery routines with an explicit behavior contract and test harness so retries and backfills don’t change outcomes.

Back to pair page

Context

Why this breaks

Oracle estates often embed business correctness in PL/SQL: packages and procedures that enforce rules, triggers that create side effects, and functions used pervasively in SQL. In migration, teams convert tables and queries first—then discover procedural logic was the real system. BigQuery can implement equivalent outcomes, but only if procedural behavior is turned into explicit routines with testable semantics for state, errors, and idempotency. Common symptoms after migration:

  • Outputs drift due to type coercion and NULL/empty-string differences - Dynamic SQL behaves differently (quoting, binding, identifier resolution) - Error handling changes; pipelines fail differently or silently continue - Trigger side effects disappear (audits, derived columns, control table updates) - Row-by-row PL/SQL patterns become expensive and fragile if ported directly A successful migration extracts the behavior contract and validates it with a replayable harness, not ad-hoc spot checks.

Approach

How conversion works

  • Inventory & classify PL/SQL assets: procedures, functions, packages, triggers, and their call sites (ETL, apps, reports). - Extract the behavior contract: inputs/outputs, side effects, error semantics, state assumptions, and performance constraints. - Choose the target form per asset:
  • BigQuery SQL UDF for pure expressions - BigQuery JavaScript UDF for complex logic where JS is appropriate - BigQuery stored procedure (SQL scripting) for multi-statement control flow - Set-based refactor where PL/SQL loops can be eliminated - Pipeline-side enforcement for trigger-like side effects (DQ, audit, derived columns) - Rewrite dynamic SQL safely using parameter binding and explicit identifier rules. - Validate with a harness: golden inputs/outputs, branch and failure-mode tests, and side-effect assertions (audit/control writes).

Coverage

Supported constructs

Representative Oracle procedural constructs we commonly migrate to BigQuery routines (exact coverage depends on your estate).

SourceTargetNotes
Oracle functions used in SQLBigQuery SQL UDFs / JavaScript UDFsPreserve typing and NULL/empty-string intent with explicit casts.
PL/SQL proceduresBigQuery stored procedures (SQL scripting)Control flow rewritten; state and side effects modeled explicitly.
Packages and shared utilitiesModular routines + shared tables/viewsReplace package state with explicit inputs and control tables.
Triggers (audit/derived columns)Pipeline-enforced side effects (procedures + DQ/audit steps)Side effects become explicit and testable.
Dynamic SQL (EXECUTE IMMEDIATE)BigQuery EXECUTE IMMEDIATE with parameter bindingNormalize identifier rules; reduce drift and injection risk.
Row-by-row procedural transformsSet-based SQL refactorsAvoid cost and reliability cliffs in BigQuery.

Compare

How workload changes

TopicOracleBigQueryNotes
Execution modelPL/SQL often relies on session state and triggersRoutines should be explicit about state, inputs, and side effectsTrigger behavior becomes pipeline steps or procedures.
Dynamic SQLString concatenation commonPrefer parameter binding and explicit identifier rulesReduces drift and injection risk.
PerformanceRow-by-row loops sometimes toleratedSet-based refactors usually required for cost/latencyProcedural loops are a frequent hidden spend risk after migration.
RestartabilityState encoded in packages/control tablesApplied-window tracking + idempotency markers enforcedRetries/backfills validated via simulations.

Examples

Examples

Illustrative patterns for moving Oracle PL/SQL behavior into BigQuery routines. Adjust datasets, types, and identifiers to match your environment.

01_udf_bigquery.sql
-- BigQuery SQL UDF example (pure expression)
CREATE OR REPLACE FUNCTION `proj.util.safe_div`(n NUMERIC, d NUMERIC) AS (
  IF(d IS NULL OR d = 0, NULL, n / d)
);
02_proc_dynamic_sql_bigquery.sql
-- BigQuery stored procedure with dynamic SQL + parameter binding
CREATE OR REPLACE PROCEDURE `proj.util.refresh_partition`(p_date DATE)
BEGIN
  DECLARE sql STRING;
  SET sql = '''
    MERGE `proj.mart.fact_orders` t
    USING (SELECT * FROM `proj.stg.orders` WHERE DATE(event_ts) = @d) s
    ON t.id = s.id
    WHEN MATCHED THEN UPDATE SET t.status = s.status, t.amount = s.amount, t.updated_at = s.event_ts
    WHEN NOT MATCHED THEN INSERT (id, status, amount, updated_at) VALUES (s.id, s.status, s.amount, s.event_ts)
  ''';

  EXECUTE IMMEDIATE sql USING p_date AS d;
END;
03_audit_side_effect.sql
-- Example side effect: explicit audit write (trigger replacement)
CREATE TABLE IF NOT EXISTS `proj.audit.proc_runs` (
  proc_name STRING,
  run_id STRING,
  started_at TIMESTAMP,
  finished_at TIMESTAMP,
  status STRING
);
04_harness_example.sql
-- Minimal harness pattern: compare routine output to golden expectations
CREATE TEMP TABLE golden AS
SELECT 10 AS n, 2 AS d, 5 AS expected UNION ALL
SELECT 10, 0, NULL;

SELECT
  n, d,
  `proj.util.safe_div`(n, d) AS got,
  expected,
  IF(`proj.util.safe_div`(n, d) IS NOT DISTINCT FROM expected, 'PASS', 'FAIL') AS verdict
FROM golden;

Workload Assessment

Migrate PL/SQL logic with a test harness

We inventory your Oracle procedures/functions/triggers, migrate a representative subset into BigQuery routines, and deliver a harness that proves parity—including side effects and rerun behavior.

Book assessment

Avoid

Common pitfalls

  • Triggers ignored: implicit side effects (audits, derived columns) disappear unless recreated.
  • Package state assumptions: session state and global variables don’t translate; state must be explicit.
  • Mixed-type branches: CASE/IF returns mixed types; BigQuery needs explicit casts to preserve intent.
  • NULL vs empty string: Oracle treats empty strings as NULL in many contexts; intent must be explicit.
  • Dynamic SQL injection risk: concatenation without bindings/escaping causes drift and risk.
  • Row-by-row loops: procedural loops should be refactored into set-based SQL to avoid cost cliffs.
  • No harness: without replayable tests, parity becomes a debate at cutover.

Proof

Validation approach

  • Compile + interface checks: each routine deploys and signatures match the contract (args/return types).
  • Golden tests: curated input sets validate outputs, including NULL/empty-string and boundary cases.
  • Branch + failure-mode coverage: expected failures (invalid inputs, missing rows) are tested.
  • Side-effect verification: assert expected writes to audit/control tables and idempotency under retries/backfills.
  • Integration replay: run routines within representative pipelines and compare downstream KPIs/aggregates.
  • Performance gate: confirm no hidden row-by-row scans; set-based refactors validated with scan bytes/runtime baselines.

Execution

Migration steps

A sequence that keeps behavior explicit, testable, and safe to cut over.

  1. 01

    Inventory PL/SQL assets and build the call graph

    Collect procedures, functions, packages, and triggers. Map call sites across ETL, applications, reports, and scheduled jobs. Identify side effects and state dependencies.

  2. 02

    Define the behavior contract

    For each asset, specify inputs/outputs, typing/NULL/empty-string intent, expected errors, side effects, restart semantics, and performance expectations. Choose the target form.

  3. 03

    Convert logic with safety patterns

    Rewrite casts and NULL/empty-string behavior explicitly, migrate dynamic SQL using bindings, and refactor row-by-row patterns into set-based SQL where feasible.

  4. 04

    Recreate trigger side effects explicitly

    Implement audit/control writes, derived column logic, and DQ enforcement as explicit steps (procedures or pipeline stages) with measurable outputs.

  5. 05

    Build a validation harness and cut over

    Create golden inputs, edge cohorts, and failure-mode tests. Validate outputs and side effects deterministically, then cut over behind gates with rollback-ready criteria.

FAQ

Frequently asked questions

Do we have to rewrite everything as BigQuery stored procedures? +

Not necessarily. Many Oracle functions can become SQL UDFs, and some PL/SQL logic should be refactored into set-based SQL. We choose the target form per asset to minimize risk and cost.

What happens to triggers? +

Triggers are a major hidden dependency. We replace trigger side effects with explicit pipeline steps (audit writes, derived columns, DQ enforcement) so behavior is visible, testable, and replayable.

How do you prove parity for procedures and functions? +

We build a replayable harness with golden inputs/outputs, branch and failure-mode coverage, and side-effect assertions. Integration replay validates downstream KPIs before cutover.

Will performance change after migration? +

It can. Row-by-row PL/SQL loops are the biggest risk. We refactor to set-based SQL where possible and gate performance with scan bytes/runtime baselines.

Migration Acceleration

Cut over routines with proof-backed sign-off

Get a conversion plan, review markers for ambiguous intent, and validation artifacts so procedural logic cutover is gated by evidence and rollback criteria.

Book assessment
Book assessment