Microsoft Dynamics GP Cloud Archive — Parquet, Queryable, Tiered

Dynamics gp cloud archive is a productised long-term data archive for Microsoft Dynamics GP — every active and historical row from your Great Plains SQL Server company databases extracted, transformed to Parquet, partitioned by company / fiscal year / module, hash-signed for tamper-evidence and landed on cloud object storage (S3, Azure Blob, GCS or OCI Object Storage) with tiered cost optimisation. A query layer (Athena, Synapse, BigQuery or Trino) and a consumer-grade search UI sit on top so finance, audit and examiner queries continue to work for the full retention window — without keeping the GP SQL Server stack alive.

A SQL Server backup is a binary blob. To answer a question from it you have to restore it to a SQL Server instance of compatible version, mount the GP databases, fire up the GP client (which still needs Dexterity runtime, Modifier resources, ISV add-ons and a valid GP licence) and run a query. That is days of work per query, costs thousands per restore and breaks every time Microsoft retires a SQL Server version. Dynamics gp cloud archive is queryable on day one and every day after — Parquet on S3 read by Athena, no SQL Server licence, no GP licence, no Dexterity runtime. Backups are still useful as cold-storage insurance; the cloud archive is what auditors actually query.

All four major clouds: AWS (S3 + Glacier tiering, Athena query, IAM access, KMS encryption), Azure (Blob Storage hot/cool/archive tiers, Synapse Serverless query, Entra ID access, customer-managed keys), GCP (Cloud Storage Standard/Nearline/Coldline tiers, BigQuery query, IAM, CMEK) and OCI (Object Storage with tiering, Autonomous Database query, IAM, customer-managed keys). The Parquet format is portable across all four — the same archive can be lifted from one cloud to another if procurement strategy changes, with no re-extraction from a long-decommissioned GP installation required.

Tiered storage means active query partitions stay on hot storage (immediate query response, higher per-GB cost) while older partitions move to cool then cold tiers (slower retrieval, dramatically cheaper per-GB). A typical small-business GP installation with 10 years of history might keep the last 18 months on hot, years 2–4 on cool, years 4–10 on cold. Total cost: USD 800–4K/year for a typical SMB footprint. Same archive on live SQL Server: USD 8K–40K/year. The tiering policy is configurable per record type — payroll might stay hot longer than inventory, for example.

Each GP company database becomes a partition in the cloud archive (company-code at the top of the partition hierarchy, then fiscal year, then module). SY01500 company master metadata is preserved as a top-level table so consumers can see the human-readable company name, address and fiscal calendar alongside the technical company code. Query patterns work three ways: query a single company (most common — auditing one entity), query across all companies (M&A diligence, consolidation reconstruction), or join across companies (inter-company transaction lookup). The Parquet partitioning makes single-company queries cheap and full-archive queries possible without exploding cost.

Yes. Three layers of evidence: (1) data-at-rest encryption with cloud-provider KMS keys or customer-managed keys, (2) tamper-evident integrity via SHA-256 hash signatures on every partition with an append-only manifest chain — any post-extract modification fails verification, (3) read-access audit log capturing every query: who, what, when, from which IP, with the log itself hash-signed. Combined, this satisfies SOX (auditable trace from GL entry to source evidence), IRS Pub 583 (substantiation produced on examiner demand with provenance), GDPR (data-subject access with read-log) and GoBD (German fiscal retention with verifiable integrity).

Yes — the cloud archive supports the full reporting pattern that finance teams ran in live GP. Trial balance per company per period, AP aging as of a historical date, AR aging as of a historical date, vendor 1099 history, customer payment history, payroll YTD per employee per tax year, fixed asset depreciation history. The critical SmartList Builder queries that finance teams relied on in live GP get rebuilt against the cloud archive — typically as parameterised SQL or BI Publisher reports — and continue to produce the same output years after GP is gone. OTBI or BI Publisher in Fusion can pull from the archive directly for blended live-plus-historical reporting.

Setup runs 4–8 weeks for a typical single-company or 2–4-company GP installation, longer for larger multi-entity rollouts. Phases: discovery and inventory (week 1), extract architecture (week 1–2), bulk extract and Parquet build (weeks 2–4), query layer and search UI configuration (weeks 3–5), UAT and retention sign-off (weeks 5–7), decommission handover (weeks 7–8). Total cost: USD 25K–80K one-time setup depending on company DB count and ISV add-on complexity, then USD 800–6K/year ongoing for storage, query and search UI. Versus keeping GP live for query at USD 8K–40K/year, payback typically arrives in year one.