Data management systems typically include a [[database]] for operational data and (optionally) a [[data warehouse]]. These support as [[OLTP]] and [[OLAP]] functions of data management, respectively. Both of these approaches are designed around a [[schema]], which requires a perspective on your data. Without this, you may consider a [[big data]] or [[NoSQL]] approach. For some applications, the [[graph database]] will be the best option. > [!Note] > See [[Introduction to databases]] for a non-technical introduction to selecting a database solution for an application. The [[relational database]] has been the dominant data management solution for decades, and still represents 70% of the databases in use today. However, relational databases can struggle to handle the requirements of modern [[big data]] applications. One option for adapting to larger volume, velocity and variety of data is scaling. Adding CPU, memory and storage is called **scaling up**. Adding server nodes to a cluster is called **scaling out**. Clustering refers to "commodity" hardware (i.e., [[cloud computing]]) but introduces some overhead for inter-node communication.