A pointer swizzling is the conversion of database objects between object identifiers (OIDs) and memory pointers. This is used to speed manipulation of memory resident objects in object-oriented database systems. In this thesis, we designed the resident object table (ROT) which represents the state of accessed objects in the client buffer manager on the basis of EXODUS storage manager (ESM). The ROT is used to manipulate the objects accessed by application. To reduce the object that was accessed by application is swapped out, the dual scanning least resently used (LRU) algorithm is added. The first phase of algorithm try to find the indirect referenced object by accessed object. If it is not exist, then the algorithm try to find least resently not used object in the second phase. This algorithm guarantees that the directly accessed objects by application are not swapped out if possible.
Our approach is different from ESM in the aspect of copy the objects into virtual memory or not. That is, the accessed objects by application are resident in buffer pool instesd of copying from the buffer pool into virtual memory. Therefore, the overhead of copying objects in ESM is reduced. In the classification of the pointer swizzling scheme, this is called lazy indirect swizzling (LIS)
