Infor LX (BPCS) Data Retention — Post-IBM i Compliance

Infor LX (BPCS) data retention is the policy and architecture for what BPCS data must be kept queryable, for how long, and on what storage substrate — after BPCS has been migrated to a new ERP and the IBM i hardware is being decommissioned. The retention question covers structured BPCS data (GL transactions, AP / AR history, sales orders, work orders, inventory transactions, audit-trail files) and IBM i journal entries that carry the SOX / FDA-relevant chain from a GL journal line back to the originating RPG program execution. Most migration projects defer the retention question to the last quarter of the project on the assumption that 'we'll just keep the IBM i running read-only' — and only then discover that running IBM i hardware in read-only mode for 7 years to satisfy SOX retention costs $200K–$1M per year and undoes most of the Power-i wind-down savings the business case was built on. Doing Infor LX (BPCS) data retention design properly up front is one of the cheapest insurance investments on the migration.

Multi-jurisdiction. United States SOX: 7 years for financial records with auditable trace from GL journal back to originating transaction (BPCS audit-trail files + IBM i journal entries). FDA 21 CFR Part 11 (pharma, medical devices): 10 years for batch records, 5 years post-distribution for distribution records, often longer per product. EU GDPR: 'no longer than necessary' for personal data — typically 6–7 years for employment-related, 10 years for tax-related, immediate deletion on right-to-erasure request for non-mandatory categories. EU CSRD sustainability: 10 years for sustainability-relevant data. UK HMRC: 6 years for VAT records. Germany GoBD: 10 years for accounting records. France: 10 years for commercial records. Asia varies — Japan 7 years for tax, China increasingly aggressive on data localisation. The Infor LX (BPCS) data retention policy has to handle the maximum window across all jurisdictions the customer operates in, typically defaulting to 10 years for the longest cohort and 7 years for the SOX baseline.

Three substrate patterns. Substrate A — Cloud object storage with hash-signed evidence: BPCS data exported as Parquet (one file per BPCS file per fiscal year per company) to S3 / GCS / Azure Blob with hash-signed manifests proving completeness. Queryable via Athena / BigQuery / Synapse / Spark SQL for retro-reporting. Lowest cost (typically $5–20 per TB per month for warm storage, $1–4 per TB per month for cold / Glacier). Substrate B — Cloud data warehouse: BPCS data loaded to Snowflake / BigQuery / Redshift / Databricks as historical tables with full BPCS schema preserved. Queryable in native SQL. Higher cost than object storage but immediate query performance. Substrate C — Read-only IBM i kept live: existing AS/400 hardware kept running in read-only mode with BPCS application running but no new transactions. Highest cost, simplest from query-perspective, undoes most Power-i wind-down savings. Most customers run Substrate A for cold history (7+ years), Substrate B for warm history (last 2–3 years), Substrate C for the first 12 months post-cutover as rollback insurance only.

The SOX / FDA chain runs from Fusion GL journal line → Fusion Journal → BPCS GLT row → BPCS GLAT audit-trail record → IBM i journal entry → original RPG program execution timestamp, user profile and terminal on the AS/400. Every link of the chain must be preserved for the retention window even after the IBM i is decommissioned. The Syntra Infor LX (BPCS) data retention pattern captures every BPCS audit-trail file (GLAT, APAT, RAAT, IIAT, MHAT) and every relevant IBM i journal entry as Parquet evidence with hash-signed manifests at extract time, routes them to cloud object storage with KMS-managed encryption and immutable storage policy (S3 Object Lock / GCS Bucket Lock / Azure Immutable Blob), and provides a query interface that lets internal audit drill from a Fusion GL journal line back through every link of the chain. Auditors can verify the chain is intact for the full retention window — typically signed off in the first post-cutover external audit cycle.

Yes, retention windows are upper bounds — data older than the longest applicable retention window can and should be deleted. The Syntra Infor LX (BPCS) data retention engine includes a retention-lifecycle scheduler that classifies every BPCS data partition by retention category (SOX 7-year, FDA 10-year, GDPR right-to-erasure-eligible, etc.), tracks expiry dates per partition, and produces deletion proposals as expiry approaches. Deletion is reviewed and signed by compliance before execution. For GDPR right-to-erasure requests on personal data (employee records, customer contact data, supplier contact data), the retention engine supports targeted deletion: per-data-subject query, deletion candidate generation, compliance sign-off, irreversible delete with hash-signed deletion evidence. The deletion evidence becomes a queryable record that 'data subject X had record Y deleted on date Z under request W'. GDPR-compliant retention is the default configuration.

Three query patterns supported. Pattern 1 — Self-service SQL: business users with appropriate access query BPCS historical data directly via Athena / BigQuery / Snowflake / Synapse with the original BPCS schema preserved and joined to Fusion lookup tables for human-readable codes (company name vs CCM code, currency description vs MCM code). Pattern 2 — Embedded in Fusion OTBI: BPCS historical data registered as a Fusion OTBI subject area via the cloud DW connector so finance can drill from a Fusion OTBI report back to underlying BPCS historical detail without leaving Fusion. Pattern 3 — Audit-trail drill: internal audit and external auditors query through a specialised drill interface that traces from Fusion GL journal back through BPCS GLT → BPCS GLAT → IBM i journal entry → original RPG program execution. Most customers run all three patterns concurrently with role-based access control via SAML / OIDC SSO.

Cost depends on substrate and data volume. Substrate A (cloud object storage with Parquet evidence): typically $500–$5,000 per TB per year all-in (warm tier) or $100–$1,000 per TB per year (cold / Glacier tier) including KMS encryption, immutable storage policy, query egress and audit logging. A typical BPCS retention footprint (10–50 TB of structured data plus 5–20 TB of audit-trail and journal evidence) costs $5K–$50K per year on the object-storage substrate. Substrate B (cloud data warehouse): typically 3–5× the object-storage cost depending on warehouse choice and query frequency. Substrate C (read-only IBM i kept live): typically $200K–$1M per year for hardware, software, MIMIX, maintenance. Most customers run Substrate A for cold (7+ year) history and Substrate B for warm (last 2–3 year) history, total typically $25K–$150K per year — vs $300K–$2M per year if the IBM i stays live.

Five governance pillars. Pillar 1 — Retention policy document signed by compliance, internal audit, finance and legal specifying retention windows per data category per jurisdiction. Pillar 2 — Immutable evidence: every BPCS record and IBM i journal entry routed to retention storage with hash-signed manifest and immutable storage policy (S3 Object Lock / equivalent). Pillar 3 — Chain-of-custody documentation: provable chain from BPCS source extraction through to retention storage with timestamped hash evidence at each step. Pillar 4 — Access control: role-based access to retention storage via SAML / OIDC SSO with audit logging of every query. Pillar 5 — Deletion governance: retention expiry workflow with compliance sign-off, hash-signed deletion evidence, and queryable deletion record. Pharma customers under FDA 21 CFR Part 11 and financial services customers under SOX routinely pass external audit of their Infor LX (BPCS) data retention on first attempt — when the five pillars are in place.