Institute for Systems Research Technical Reports

In this paper we compare the three materializationpolicies (materialized inside the DBMS, materialized at the web serverand virtual) analytically, through a detailed cost model, andquantitatively, through extensive experiments on an implementedsystem. Our results indicate that materializing at the web server is amore scalable solution and can facilitate an order of magnitude moreusers than the virtual and materialized inside the DBMS policies, evenunder high update workloads.

For the materialized case, updates to base data lead to immediate recomputation of the WebView, whereas in the virtual case, recomputation is done on demand with each request. We introduce the materialize on-demand approach which combines the two strategies, and generates WebViews on demand, but also stores the results and re-uses them inthe future if possible.

Deciding on one of the three materialization policies for each WebView is clearly a performance issue. In this paper, we give the framework for the problem and provide a cost model, which we test with experiments on a real web server.

The value of the network management data collected diminishes as the data ages. The value that remains is in the trend that the data outlinesof the macroscopic behavior of the network. This thesis proposes an algorithm which highlight these trends and, in the process, causes a significant data reduction. The proposed algorithm's reductionmethodology is compared with standard data reduction techniques. It isfurther shown that the processed data is suited for storage in a temporaldatabase. The benefits of storing time series data, such as the networkmanagement data discussed (i.e., Link Utilization, CPU Load, and DeviceReachability), in a temporal database is demonstrated by storing the data in both a rational Oracle 7 and a temporal TIGER database, and examining results of queries against both databases.

In this paper, we describe our current work towards the integration and enhancement of the capabilities of EDAPS and EXTRA. We integrate EXTRA's functionality with the initial design step of EDAPS. in the resultant system, the user, supported by an enhanced tradeoff analysis capability, can select and describe a promising preliminary design and process plan based on the analysis of a variety of alternatives from both an electrical and mechanical perspective. This preliminary design is then subjected top further analysis and refinement using existing EDAPS capabilities. In addition to the integration of these two systems, specific new functions have been developed, including tradeoff analysis over a much broader set of criteria, and the ability of the tradeoff module to query the process planner to determine costs of individual options.

Specifically, we used state-of-the-art concepts from morphology, namely the attern spectrum' of a shape, to map each shape to a point in n-dimensional space. Following [19, 36], we organized the n-d points in an R-tree. We showed that the L (= max) norm in the n-d space lower-bounds the actual distance. This guarantees no false dismissals for range queries. In addition, we developed a nearest neighbor algorithm that also guarantees no false dismissals.

Finally, we implemented the method, and we tested it against a testbed of realistic tumor shapes, using an established tumor-growth model of Murray Eden [15]. The experiments showed that our method is up to 27 times faster than straightforward sequential scanning.

The problem of concurrency control and scheduling of transactions in real time database systems is studied in the framework of discrete event dynamical systems (DEDS) modeled by deterministic finite automata (DFAs). Concurrency control and scheduling are separated into two different modules (a logical DEDS model for the CC module and a heuristic implementation of a scheduler) to allow modular analysis of various combinations of concurrency control and scheduling algorithms. The model is developed analytically using the theory of discrete event dynamical systems. Subsequently the design of a simulation software is reported that uses this model to simulate transaction execution for a (concurrency controller, scheduler) pair. Finally, we show that our approach can also be viewed as a special case of a supervisory control theory (SCT) synthesis technique. The goal of this thesis is to demonstrate the applicability of DEDS theory as a powerful tool in modeling and analyzing transaction models in real time database systems and to show potential applications of modern SCT techniques in this area.