7 Questions To Ask When Choosing Your Data Architecture
Choosing the right enterprise data architecture can be confusing. Our infographic sets out 7 key questions you need to ask to help make the right choice for your business.
Should you opt for a data warehouse? Or maybe a data lake or a data vault is a better option? Here are 7 questions to guide you through the factors you should consider in order to ensure your data storage solution works for your needs.Infographic: 7 Questions When Choosing Your Data Architecture
1. What initial investment cost do you want?
If you’re looking for a low initial investment, consider an operational data architecture — where no initial setup is required, or an operational data store (ODS), where no excessive modelling is required. Cost of initial investment is slightly higher for data lakes, where most initial modelling can be done later, or a data mart, offering easier modelling. However, if high initial investment is is an option for you, consider a data vault or a data warehouse.
2. What cost are you comfortable with when it comes to asking new questions?
If you’re looking for a low cost when new questions arise, consider a data vault or a data lake architecture, or even an ODS. Data marts and operational data architectures incur slightly higher costs when new questions arise, and asking new questions within a data warehouse architecture incurs the highest cost as it requires remodelling of the global normalized model.
3. What are your ongoing maintenance costs?
Data marts, data warehouses and operational data architectures offer low or no ongoing maintenance costs. However, if you are considering a data vault or data lake, changes in data within data vaults and storage costs of data lakes require higher ongoing maintenance costs.
4. How often does your business change?
If there are constant changes in the business, data lakes and operational data architectures are best suited to this kind of environment. If the change is slow, then a data warehouse or data mart architecture could work well in this environment. A data vault tends to respond well in a medium change frequency environment.
5. What type of data do you have?
Is your data highly structured? A data warehouse, data mart or data vault architecture are suitable for this type of data. However, if it’s mixed with unstructured data, a data lake or operational data architecture are worth considering.
6. What's your use case?
Above all, your use case is fundamental to the data architecture decision-making.
- If your use case is...BI, KPI Reporting
- You should consider... a data warehouse or a data mart
- If your use case is...Ad hoc analytics on structured data
- You should consider... a data lake or a data vault
- If your use case is... Ad hoc analytics on any data
- You should consider... a data lake or operational data
- If your use case is... Near real-time fresh data required
- You should consider...operational data or an ODS
7. What talent do you have available or planned in the organization?
At the lower end of the scale, non-technical talent suits a data warehouse or data mart architecture, whereas at the other end of the scale, data lake or operational data architectures suit the more data-savvy talent, such as data scientists, developers or data consultants. Where the more traditional IT talent exists, data warehouse, data mart and data vault architectures are worth considering.
If you want to discover more about your options for enterprise data architecture, watch our webinar Choosing the Right Data Architecture for your BusinessData warehouses, lakes, hubs, and vaults explained