Real-Time Databases and Data Services
Real-Time Systems
Distributed real time database systems: background and literature review
Distributed and Parallel Databases
| Hi-index | 0.00 |
EVEN THOUGH CONSIDERABLE RESEARCH HAS BEEN DONE ON OCNCURRENCY CONTROL PROTOCOLS THAT TAKE THE TIMING CONSTRAINTS OF TRANSACTIONS INTO ACCOUNT, MOST OF THESE PROTOCOLS DO NOT PREDICT -- BEFORE A TRANSACTION BEGINS EXECUTION -- WHETHER THE TRANSACTION WILKL MEET ITS DEADLINE. HENCE THE PROTOCOLS DO NOT DIRECTLY TACKLE THE PROBLEMS INTRODUCED BY THE VARIOUS SOURCES OF UNPREDICTABLITY ENCOUNTERED IN TYPICAL DATABASE SYSTEMS, IN- CLUDING DATA DEPENDENCE OF TRANSACTION EXECTUTION, DATA AND RESOURCE CONFLICTS, DYNAMIC PAGING IN VIRTUAL MEMORY SYSTEMS, DISK I/O, AND TRANS- ACTION ABOURTS WITH THE RESULTING ROLLBACKS AND RESTARTS. ON THE OTHER HAND, THE APPROACK DESCRIBED IN THIS PAPER HAS THE POTENTIAL TO PROVIDE PREDICTABLE TRANSACTION EXECUTIONS. THIS APPROACH EXPLOITS THE ACCESS INVARIANCE PROPERTY THAT MANY TRANS- ACTIONS POSSESS OR CAN BE DESIGNED TO POSSESS. THE EXECUTION PATH OF A TRANSACTION WITH THIS PROPERTY IS UNLIKELY TO CHANGE AS A RESULT OF DATA MODIFICATIONS BY OTHER CONCURRENT TRANSACTIONS. IN THE FIRST PHASE, CALLED THE PREFETCH PHASE, A TRANSACTION IS RUN ONCE, BRINGING THE NECESSARY DATA INTO MAIN MEMORY THAT IS NOT IN MEMORY ALREADY. NO WRITES ARE PERFORMED IN THIS PHASE AND DATA CONFILCTS WITH OTHER TRANSACTIONS ARE NOT CONSIDERED. THE OVERALL COMPUTATIONAL DEMANDS OF THE TRANSACTION ARE ALSO DETERMINED DURING THIS PHASE. AT THE END OF THE PREFETCH PHASE, THE SYSTEM ATTEMPTS CONTINUED... TO GUARANTEE THAT THE TRANSACTION WILL COMPLETE BY ITS DEADLINE. THIS IS DONE BY PLANNING THE EXECUTION OT THE TRANSACTION -- TAKING INTO ACCOUNT DATA AND RESOURCE CONFLICTS WITH THE TRANSACTIONS ALREADY GUARANTEED -- SUCH THAT THE TRANSACTION MEETS ITS DEADLINE. IF SUCH A PLAN CAN BE CON- STRUCTED, THE TRANSACTION''S EXECUTION PHASE BEGINS AND IF ACCESS INVARIANCE HOLDS THEN THE TRANSACTION IS COMMITTED AT THE END OF THIS PHASE. A NUMBER OF VARIATIONS ON THIS APPROACH (SUCH AS USING OPTIMISTIC CONCURRENCY CON- TRPL IN THE PREFETCH PHASE OR HANDLING OCCAISIONAL FAILURES OF ACCESS IN- VARIANCE) ARE ALSO INVESTIGATED. THE BENEFITS OF THE APPROACH ARE ALSO DISCUSSED.