The Duck Creek Cloud Archive — Parquet on Object Storage, Insurance-Governed

The duck creek cloud archive is a Parquet-on-object-storage product that holds Duck Creek Policy, Billing, Claims, Treaty and statutory reporting historical data in immutable, indexed, queryable form on S3 / GCS / Azure Blob / OCI Object Storage — with tiered storage routing per record age and per-state retention policy enforcement. On-prem retention typically means keeping the source Duck Creek Platform alive on premises with its SQL Server or Oracle DB stack, or exporting flat-file dumps to NAS. Both approaches preserve the bytes but make retrieval slow and governance ad-hoc. The duck creek cloud archive is queryable in sub-15 seconds per record, immutable via S3 Object Lock, governed with audit-grade access logging, and 70–85% cheaper at scale than the on-prem alternative.

Three reasons. First, Parquet is columnar — analytical queries that scan one column across millions of historical policy or claim records run 10–100× faster than scanning the equivalent SQL Server or Oracle DB row-store. Second, Parquet is open and self-describing — state insurance department examiners, external auditors, reinsurance counterparty auditors and successor-system vendors can read it without licensing SQL Server, Oracle DB or the original Duck Creek Platform application stack. Third, Parquet integrates natively with the analytic engines users already have (Snowflake, Redshift, BigQuery, Databricks, Spark, Oracle Autonomous Database, DuckDB) — the duck creek cloud archive plugs into your data platform without an intermediate ETL layer. The original Duck Creek source-format data is preserved as a checksum-validated backup alongside the Parquet for byte-level reconstruction if ever required.

AWS, Google Cloud, Azure and OCI for customers consolidating on Oracle Cloud Infrastructure alongside Fusion. Storage tiers per provider: AWS S3 Standard / S3 Infrequent Access / S3 Glacier Instant Retrieval / S3 Glacier Flexible Retrieval / S3 Glacier Deep Archive, with S3 Object Lock enforcing immutability. GCS Standard / Nearline / Coldline / Archive, with Bucket Lock for immutability. Azure Blob Hot / Cool / Cold / Archive, with immutable blob storage. OCI Object Storage Standard / Infrequent Access / Archive with WORM retention rules. Tiered routing happens automatically per record age — recent retention (years 0–3) on the hot tier, mid-window (years 3–10) on infrequent-access, long-tail (years 10–30+) on Glacier / Archive. Retrieval from Glacier tiers takes minutes; from hot tiers, milliseconds.

Every cloud provider supported under the duck creek cloud archive product has a published compliant configuration with appropriate contracts signed at the storage account level. KMS encryption keys live in your account, rotated per your policy. The per-state retention engine carries each state's insurance retention rule per line of business per record type — California 6 years personal lines and 10+ workers' comp, New York 6 years post-termination, Texas 5 years policies and longer for claims, Florida 5 for property and 10+ for medical malpractice. Long-tail liability claims (asbestos, environmental, construction defect) get statute-of-limitations extensions. S3 Object Lock (or equivalent) enforces the computed window. Records purged past every applicable retention generate a signed certificate of destruction. State examiners see one consolidated retention story per record.

Yes — through role-specific interfaces sharing one underlying archive. Statutory accountants retrieving prior-year Annual Statement reconciliation access through OTBI / BI Publisher with NAIC-aware subject areas. Litigation counsel retrieving subpoena, coverage dispute and bad-faith claim records accesses through the litigation discovery interface scoped to the policy or claim at issue. Reinsurance treaty auditors retrieving per-treaty per-bordereau cession history access through the treaty audit interface scoped to the relevant treaty period. State market-conduct examiners conducting periodic exams access through the exam portal with engagement-scoped access. Internal audit retrieves through their own portal. Every retrieval logged centrally for audit substantiation.

Storage cost depends on data volume and tier mix. A typical mid-market carrier retiring a 20-TB Duck Creek Platform instance with a 30-year retention window (3 years hot, 7 years infrequent-access, 20 years Glacier) costs $40K–$120K per year in storage. A multi-line P&C carrier retiring 100+ TB across multiple legacy Platform instances after M&A consolidation costs $200K–$700K per year in storage. On top of storage, query and operational infrastructure adds $15K–$80K per year depending on query volume. Total typical TCO: $55K–$780K per year depending on scale, vs $500K–$5M+ per year for keeping the equivalent source Platform instances live. The 70–85% TCO reduction reported by carriers migrating to the duck creek cloud archive is consistent across scale.

When a P&C carrier consolidates finance onto Oracle Fusion (with Duck Creek continuing to run insurance operations through OnDemand and/or Platform), the duck creek cloud archive becomes the long-term source of historical insurance-finance truth. Fusion's analytic layer (OTBI, BI Publisher) registers the archive Parquet store as an external data source via Oracle Autonomous Database or Parquet connector — a single OTBI dashboard combines current Fusion data with historical Duck Creek data without ETL'ing the archive into Fusion. NAIC Annual Statement reconstruction continues working against the archive. Reinsurance Schedule F reconstruction continues working against the archive. The cloud archive sits invisibly behind the analytic surface that statutory accountants, finance and actuarial all use.

Legal holds are per-policy, per-claim or per-litigation overrides that extend retention indefinitely while active. The claims litigation team places a hold via the archive admin interface, scoped to policy identifier, claim identifier, claimant identifier, date range or matter ID. Affected records' retention windows extend until the hold is released. Discovery requests against held records run as standard archive queries scoped to the matter, with results delivered in the format requested (PDF package, native record export, structured ESI for opposing counsel). Every discovery retrieval is logged centrally for audit substantiation and for the litigation audit trail. When the hold is released, retention recomputes from the original per-state rules and records past expiration become eligible for purge per the standard certificate-of-destruction workflow. Long-tail liability claims (asbestos, environmental, construction defect) effectively stay on legal hold for decades because the underlying litigation never closes.