Choose around the workload.
Architecture begins with how data is created, accessed, protected and retained.
| Architecture | Typical workload | Questions to ask |
|---|---|---|
| Block | Databases and virtual machines | What latency, consistency and availability does the application require? |
| File | Shared documents, engineering and media | Which protocols, permissions and concurrent access patterns matter? |
| Object | Unstructured data, archives and data lakes | Which APIs, metadata, retrieval times and lifecycle policies fit? |
| Cloud storage | Region-based managed storage services | What are the location, transfer, request and operating requirements? |
These are starting points. A deployment can combine several models; cloud describes a delivery environment, not a single access protocol. Object-storage reference →
Performance is more than a capacity number.
Measure the workload before choosing hardware or service tiers.
- Latency: response time, including peak periods and tail behavior.
- Throughput: sustained data transfer under representative concurrency.
- IOPS: operation size and read/write mix matter as much as the count.
Location is an operating decision.
Evaluate the whole path between data, people and applications.
- Classify residency, jurisdiction and administrative access requirements.
- Model egress, retrieval and API request costs using supplier quotes.
- Test portability and migration windows before committing to a tier.
A copy is only useful if you can recover it.
Recovery needs an operational design, not just additional terabytes.
- Set the recovery point objective (acceptable data loss) and recovery time objective (acceptable outage).
- Consider isolated credentials, immutable retention and independently protected copies.
- Run restore exercises and verify application consistency; capacity alone cannot prove resilience.
FROM INSIGHT TO A PLANStart with your data.
Open capacity planner →