The CORDS multidatabase project

Abstract

AlTALURI ET AL. 43 *Trademark or registered trademark of International Business Machines Corporation.

Evolutionary techniques for updating query cost models in a dynamic multidatabase environment

“…Assume the sample size for multiple regression analysis in the query sampling method is k. 2 At time t k−1 , we have k sample data points…”

“…In other words, we should incorporate the performance information contained in the newest sample data point (y k , x T k ) into the cost model since it reflects the current system environment. On the other hand, we also need to remove the oldest sample data point (y 0 , x T 0 ) from consideration for the cost model since (i) it contains the least information about the performance behavior of the current system environment and (ii) keeping all old sample data points would make the sample set size grow big- 2 A commonly used rule for sampling is to sample at least ten observations for every parameter to be estimated [13], i.e., at least 10 * (n + 2) sample queries in our case, where n is the number of explanatory variables in the cost model (note that the variance of error terms is also one parameter to be estimated here).…”

“…may be small, assuming the environment changes slowly, the difference between two far-away cost models M (i) and M (i+q) , where q is large, can be very significant. Using the up-to-date cost models M (0) , M (1) , M (2) , . .…”

Developing cost models with qualitative variables for dynamic multidatabase environments

“…To verify the feasibility of our multi-states query sampling method for developing cost models in a dynamic environment, experiments were conducted in a multidatabase environment using a research prototype called CORDS-MDBS [2]. Two commercial DBMSs, i.e., Oracle v8.0 and DB2 v5.0, were used as local database systems running under Solaris 5.1 on two SUN UltraSparc 2 workstations.…”

“…Their observed data as well as their associated probing query costs are recorded and used to derive a regression cost model for the query class. A load builder, which is part of the MDBS agent for each local DBS [2], is used to simulate a dynamic application environment at a local site in an MDBS during the query sampling procedure. The MDBS agent may also have an environment monitor which collects system statistics used for estimating the probing query costs when the estimation approach in Section 3.3 is employed.…”