Database Management System (DBMS) workload involves homogenous as well as heterogeneous data and concurrent users. Humans are incapable to manage the versatile data and dynamic behavior of DBMS workload. There is a need of fast computations of current server’s loads and requirements, AI algorithms and machine learning techniques. Autonomic computing technology using types of workload Decision Support System (DSS) or Online Transaction Processing (OLTP) and its performance requirement can help servers, adaptation of the workloads. If we know the type of workload, we can design such systems that predict the identified workload performance and adapt the changes in the behavior of the workload. For managing the workload, we have to face number of problems for the DBMS to better perform. Before executing, we can predict and control the workload to tune the DBMS. Predicting performance of the workload is important for tuning a DBMS and makes the DBMS aware of itself making it autonomic. The optimizer and DBMS can tune itself accordingly. Evolving behavior of workload can be handled by making the system adaptive. We have developed a framework called Autonomic Workload Performance Predictor (AWPP) for predicting the performance of the workload making it adaptive to the changing behavior of the workload. Case-based reasoning approach is applied and results are compared with other well-known machine learning techniques to observe the accuracy and effectiveness and significance of AWPP framework. MySQL database management system is being used to execute different benchmark workload to validate the proposed workload performance prediction framework. For training and testing TPC- H and TPC-C like queries are used as our representative workload. We have taken the various benchmark workloads of DSS and OLTP for experimentation. CBR approach produced effective, accurate and significant results while predicting the performance of workload using the information available before executing a workload and adapting the workload on evolution. These predictions will be helpful for optimizer and DBMSs algorithms as well as for workload management, capacity planning, system sizing.