Workload Requirements
Document I/O profile for each workload
IOPS (read/write ratio), throughput (MB/s), latency requirement, and block size. Use storage performance monitoring tools to capture actual I/O profiles, not estimates.
Identify latency-sensitive workloads
Workloads requiring sub-millisecond latency require NVMe all-flash storage. Identify these workloads before evaluating storage options.
Document AI/ML storage requirements
Training dataset size, sequential read throughput required, and whether NVMe-oF is required for shared access.
Identify compliance requirements per workload
HIPAA, PCI DSS, FedRAMP: identify all applicable frameworks and their specific storage requirements (encryption, retention, audit logging).
Capacity Planning
Document current data footprint
Total data currently stored, including all copies, snapshots, and replicas. Use actual measurements, not estimates.
Document data growth rate
Historical growth rate for the last 12–24 months. Adjust for planned new workloads, particularly AI datasets.
Verify data reduction ratios
Run a proof-of-concept with actual production data to verify the data reduction ratio the array will achieve. Do not use vendor-claimed ratios.
Size for 80% utilization at end of lifecycle
Storage arrays perform best at 70–80% utilization. Size for 80% utilization at the end of the planned 5-year lifecycle.
Data Protection Design
Define RPO and RTO per workload
Recovery Point Objective (how much data loss is acceptable) and Recovery Time Objective (how long recovery can take) for each workload tier.
Select RAID/erasure coding scheme
RAID 6 minimum for enterprise arrays. Verify that the selected scheme meets the RPO requirements for the workloads.
Design replication architecture
Synchronous replication for RPO=0 requirements. Asynchronous replication for workloads tolerating some data loss. Define replication targets and network requirements.
Design snapshot schedule
Snapshot frequency and retention period for each workload. Verify that snapshot space consumption is included in capacity planning.
Compliance Requirements
Verify encryption at rest capability
Confirm that the selected array supports encryption at rest with the required algorithm (AES-256) and key management model (self-encrypting drives, external key manager).
Verify data residency requirements
For regulated data, confirm that the storage system can enforce data residency, preventing data from being replicated to locations that violate data sovereignty requirements.
Verify retention policy enforcement
For compliance workloads, confirm that the storage system can enforce retention policies, preventing data from being deleted before the required retention period.
Verify audit logging
Confirm that the storage system provides audit logs for all data access and administrative operations, with the retention period required by applicable compliance frameworks.
Migration Planning
Document current storage inventory
Complete inventory of all storage systems being replaced: including data volumes, protocols, and the applications that depend on each.
Define migration sequence
The order in which data will be migrated to the new storage system. Migrate lower-criticality workloads first to validate the migration process.
Define rollback procedures
How will the organization revert to the legacy storage system if a migration fails? Rollback procedures must be documented and tested before migration begins.
Plan maintenance windows
Maintenance windows for each migration event, agreed with the business units whose operations will be affected.
PoC testing is a planning step, not a procurement step