Law Firm Cloud Migration With Litigation Holds and Retrieval SLA
Halverson & Reed, a 180-attorney litigation firm, is moving 8 TB of scanned case files to a cloud repository. The files span 15 years and include imaged pleadings, deposition exhibits, and client correspondence. Retention is 15 years after matter closure, but 42 active matters are under litigation hold and must be preserved immutably. Attorneys expect retrieval by client, matter, document type, and date within 15 minutes; paralegals currently search by folder names that vary by office. The firm also needs to prove that no stored image has changed since capture. The proposed cloud tier costs $0.023 per GB per month for primary storage and $0.004 for archive, with egress fees after 1 TB. The records manager must design the metadata schema, storage tiers, and integrity checks without breaking the retrieval SLA or the legal hold.
1. Design the metadata schema and indexing approach. Which fields are mandatory, and how do you handle inconsistent legacy folder names?
Read the worked answer
Mandatory fields: client ID, matter ID, document type from a controlled vocabulary, document date, capture date, source office, batch ID, page count, and retention code. Add optional fields for author, recipient, and exhibit number. Use a unique document ID generated at capture and never reused. For legacy inconsistencies, do not rewrite folder names; map them to the schema with a crosswalk table and keep the original path as a searchable attribute. Normalize dates to ISO 8601 and party names to an authority list during migration. Store metadata in a database index separate from the object store so retrieval by client, matter, type, and date can use compound indexes and return in seconds, not minutes. Full-text OCR should be a secondary index with confidence scores. Require that every object has a checksum and that the metadata record is written in the same transaction as the object, or reconciled nightly.
2. How would you tier the 8 TB across primary and archive storage while meeting the 15-minute retrieval SLA and controlling egress costs?
Read the worked answer
Put active matters and any matter under litigation hold in primary object storage; that is a small fraction of 8 TB, so the cost is manageable. Move closed matters older than two years to archive tier, but keep metadata and a searchable index in the hot database so attorneys can find the document without touching archive. Use lifecycle rules to transition objects automatically and to restore on demand. For the 15-minute SLA, pre-warm the archive by restoring at the folder or matter level, not page by page; cache restored objects for 30 days. Monitor egress: set a 1 TB monthly budget alert and route bulk exports through a staging area with compression. Because egress fees apply after 1 TB, avoid emailing large sets; use a secure link that streams from the repository. This keeps 90 percent of retrieval requests in primary or cache while archive holds the cold bulk.
3. What integrity and legal-hold controls prove that no stored image has changed?
Read the worked answer
Compute a SHA-256 hash at capture and store it in an immutable audit log separate from the object. Recompute hashes on a scheduled scrub, say 1 percent of objects monthly and 100 percent before any production, and alert on mismatch. Enable object lock or WORM retention on the archive for the 42 matters under hold; apply a legal-hold flag in metadata that prevents deletion and blocks lifecycle transition to a deletable tier. Log every read, download, and metadata change with user, timestamp, and IP. For hold release, require dual approval and retain the audit trail for the full 15-year period. Do not rely on the cloud provider's default versioning alone; configure versioning plus MFA delete. Produce a certificate of integrity for any matter that goes to court, showing hash values and the chain of custody from scanner to repository.