Managing frequent updates is one of most importantissues in many update-intensive applications, e.g.,location-aware services, and stream databases. In thispaper, we present an R-tree-based index structure whichemploys semi-bulkloading (SBL) technique forefficiently managing frequent updates from massivemoving objects. The basic idea of SBL is to buffer theincoming updates in main-memory buffer, choose aproper subset from the buffer, and then bulk-insert themat once. For this purpose, we devise an efficient updatebuffer management scheme which provides an effectiveway to manage the incoming updates in memoryefficientmanner. Our experimental results reveal that theproposed approach is far more efficient than previousapproaches for managing frequent updates under varioussettings.
展开▼