The ability of a database management system (DBMS) to detect problems or problem trends (which have not yet manifested as problems) and either take corrective actions, and/or provide advice or an automated notification to the database administrator (DBA) is absolutely crucial to ensure the "always-on" and "always-operational" environments. Any DBMS related failure could be prohibitively expensive and extremely damaging to business prospects. Today's database management systems primarily rely on DBAs to manually take corrective actions to minimize failures and provide robust platform to run applications. This approach is not very efficient, expensive and often not feasible. The goal of this paper is to outline several research directions and approaches that will lead to scalable and efficient self-healing database management systems. In this paper, we propose a general framework design, termed the self-healing framework (SHF), that can enable automatic problem detection and problem resolution inside the database server, thus transforming the DBMS to become self-healing.