Factors:
- Based on different use cases.
- We can use Data Marts specifically built for that use case. A financial-related data mart cannot be used for a user session analytical.
- Use cases:
- Transactional
- Analytical
- Archival
- Data Warehousing
- Compability to the ecosystem.
- If the whole system is using AWS’ ecosystem, it is a better idea to use AWS technology than Microsoft’s.
- Types of data: structured, semi-structured, unstructured.
- For unstructured data, we cannot use RDBMS.
- Is schema predefined?
- If the schema is predefined, we should use a Data Warehouse and RDBMS rather than a Data Lake with NoSQL.
- Performance requirements, storage requirements, security requirements.
- Whether working with data at rest or streaming data
- Streaming data like time-series requires Column-based databases for efficient access and process.
- Frequency of data access, and backup options.
- Lower cost data Repository is preferred for low-frequent access data as well as unimportant data.
- Standards of the organization.
- The organization may prefer using open-source technologies to closed-source technologies.
Khiem Ton That