Modern enterprise applications are a challenge for traditional database systems. In the recent years, enterprise workloads evolved from simple lookups and updates to complex queries integrating both analytical and transactional data. In addition, graph structures and queries today need to be integrated for applications like materials requirements planning, which add some further complexity to the workload. We hence observe new challenges for enterprise database systems: To improve the effectiveness of business processes, it becomes necessary to semantically enrich business data with additional graph data. However, in order to achieve the best performance, we cannot store graph data in a traditional relational database management system. In this vision paper, we describe a new database storage architecture that is able to work with multiple storage types like traditional relations stored in rows, columns, hybrid partitions, and that is able to store graphs natively. In-Memory technology is the technological foundation of this architecture, since it provides the flexibility and performance to combine those different storage types into a single storage engine without significant performance penalty.