Self-Prediction of Performance Metrics for the Database Management System Workload

Workload in Database Management System (DBMS) consists of huge amount of data and number of concurrent users who are executing different requests that require some resources. To manage these types of activities, organizations hire different database experts. There is versatility in workload due to the huge data size and different types of requests (workload). These factors contribute to some new challenges in the workload management. These challenges are identification of the workload and decision about the problem queries, identification of resource oriented and contention queries, accurate workload classification, optimal plan selection, prediction and adoption. In DBMS, where workload management and tuning is performed through if-then approach, unforeseen behavior of the workload cannot be handled and sometime leads to unpredictable state. In this research a prediction framework has been proposed called as workload queries performance Predictor. The predictor will predict the performance metrics (workload size, elapsed time, record accessed, record used, disk I/Os, memory required, message count and bytes) for queries in a given workload. We are improving efficiency and reducing search time when projection of query feature vector is performed over performance feature vector. The predictor will take help from the optimizer and store the information in database which saves the information as history for the future.

[1]  Chetan Gupta,et al.  PQR: Predicting Query Execution Times for Autonomous Workload Management , 2008, 2008 International Conference on Autonomic Computing.

[2]  Jeffrey O. Kephart,et al.  An architectural approach to autonomic computing , 2004, International Conference on Autonomic Computing, 2004. Proceedings..

[3]  Hongjun Lu,et al.  Dynamic Task Allocation in a Distributed Database System , 1985, ICDCS.

[4]  Archana Ganapathi,et al.  Predicting Multiple Metrics for Queries: Better Decisions Enabled by Machine Learning , 2009, 2009 IEEE 25th International Conference on Data Engineering.

[5]  Gregory R. Ganger,et al.  Observer: keeping system models from becoming obsolete , 2007 .

[6]  Kevin Wilkinson,et al.  Managing operational business intelligence workloads , 2009, OPSR.

[7]  Gregory R. Ganger,et al.  Ursa minor: versatile cluster-based storage , 2005, FAST'05.

[8]  Gail E. Kaiser,et al.  Self-managing systems: a control theory foundation , 2005, 12th IEEE International Conference and Workshops on the Engineering of Computer-Based Systems (ECBS'05).

[9]  Jana Koehler,et al.  On Autonomic Computing Architectures , 2003 .

[10]  Gregory R. Ganger,et al.  Towards Self-Predicting Systems: What If You Could Ask "What-If"? , 2005, 16th International Workshop on Database and Expert Systems Applications (DEXA'05).

[11]  Anastasia Ailamaki,et al.  Continuous resource monitoring for self-predicting DBMS , 2005, 13th IEEE International Symposium on Modeling, Analysis, and Simulation of Computer and Telecommunication Systems.

[12]  David J. DeWitt,et al.  Dynamic Memory Allocation for Multiple-Query Workloads , 1993, VLDB.

[13]  Chetan Gupta,et al.  Automatic Workload Management for Enterprise Data Warehouses , 2008, IEEE Data Eng. Bull..

[14]  Said Elnaffar,et al.  An Intelligent Framework for Predicting Shifts in the Workloads of Autonomic Database Management Systems , 2004 .

[15]  Gerhard Weikum,et al.  Performance Evaluation of an Adaptive and Robust Load Control Method for the Avoidance of Data-Contention Thrashing , 1992, VLDB.

[16]  Said Elnaffar,et al.  Automatically classifying database workloads , 2002, CIKM '02.

[17]  Sam Lightstone,et al.  Toward autonomic computing with DB2 universal database , 2002, SGMD.

[18]  Said Elnaffar,et al.  A methodology for auto-recognizing DBMS workloads , 2002, CASCON.

[19]  Salim Hariri,et al.  Autonomic Computing: An Overview , 2004, UPP.

[20]  C. Hardouin,et al.  Highly enantioselective hydrogenation of new 2-functionalized quinoline derivatives , 2012 .

[21]  Harumi A. Kuno,et al.  Quality of Service-enabled Management of Database Workloads , 2008, IEEE Data Eng. Bull..