Conclusion
Self tuning technique presented in this paper, using the
learning ability of the neural network and ability to act on
imprecise data of the fuzzy logic shows significant
improvement in the response time as compared to autotuning
feature of the modern DBMS. Moreover, the ability of
the proposed method to generate almost flat response time as
compared to the auto-tuned method with increasing user load
enables the DBA to implement system that have stringent
response time requirements. The novel technique of moderating
the value of the estimated parameter based on Impact
factor avoids over tuning and thus preserves memory that
could be used for more productive purposes. However, further
research is required to establish similar facts in other
DBMS and also for different workload types. Furthermore,
the impact of one tuning parameter on the other tuning
parameter is to be measured and incorporated in the impact
factor, so that moderation step can further improve in its
action to limit the values of the tuning parameters.
Acknowledgments We deeply acknowledge the support in the form
of computing facilities and funding from our esteemed management
of Karnataka Law Society, Belgaum, Karnataka. Our thanks are also
due to our Principal, Dr. A.S. Deshpande for his support and
encouragement. We also acknowledge the contributions of Mr. Sumeet
of VIIIth semester B.E., Information Science and Engineering
department for his assistance in setting up the laboratory for carrying
out the experiments related to this research work. Our thanks are also
due to Mr. Moogbasav, Instructor, Computer Center, GIT, for providing
us with the necessary support in setting up of the test bed for
the experiments.