Classify the data before choosing tools
Structured data follows a consistent model. Semi-structured data contains organizing markers without requiring every record to share one fixed layout. Unstructured data does not naturally fit rows, columns, or key-value fields.
- Relational tables are structured.
- JSON event records are commonly semi-structured.
- Audio, video, and free-form documents are commonly unstructured.
Separate operational and analytical needs
Operational systems support current transactions and frequent small changes. Analytical environments support historical comparison, larger reads, and decision-making. The same organization may use both and move prepared data between them.
- Transaction systems prioritize reliable day-to-day updates.
- Warehouses organize curated data for repeatable analysis.
- Lakes retain raw data in multiple formats for later processing.
Compare deployment environments
Cloud services can offer rapid scaling and managed capabilities. On-premises systems provide direct local control. Hybrid designs combine environments when workloads have different performance, security, regulatory, or cost requirements.
- Elasticity is useful when demand changes.
- Availability requirements influence architecture.
- Governance obligations continue regardless of deployment model.
Quick Reference
Important Terms
- Schema
- A formal description of how data is organized, including fields, types, and relationships.
- Semi-structured data
- Data with self-describing keys or markers that does not require every record to use an identical layout.
- OLTP
- A workload pattern designed for frequent, short transactions that insert or update operational records.
- Data warehouse
- An integrated analytical repository containing prepared, historically useful data for consistent reporting.
- Elasticity
- The ability to adjust computing resources as workload demand increases or decreases.
Original Public Practice
Check Your Understanding
1. A system must record individual card transactions immediately throughout the day. Which workload pattern is the closest fit?
Show answer and explanation
Answer: Online transaction processing (OLTP).
OLTP supports frequent, short operational transactions such as individual purchases.
2. A collection contains JSON documents with different optional fields. How is the collection usually classified?
Show answer and explanation
Answer: Semi-structured data.
JSON uses keys and organization but does not require every document to share one fixed schema.
3. An organization uses local systems for sensitive records and cloud capacity for temporary analytical demand. What deployment model is this?
Show answer and explanation
Answer: A hybrid environment.
Hybrid designs combine on-premises and cloud resources within the overall solution.