This paper presents an active database discrimination network algorithm called Gator. Gator is a generalization of the widely-known Rete and TREAT algorithms. Gator pattern matching is explained, and it is shown how a discrimination network can speed up condition testing for multi-table triggers. The structure of a Gator network optimizer is described, along with a discussion of optimizer performance, output quality, and accuracy. This optimizer can choose an efficient Gator network for testing the conditions of a set of triggers, given information about the structure of the triggers, database size, attribute cardinality, and update frequency distribution. The optimizer uses a randomized strategy to deal with the problem of a large search space. Optimizer validation was performed, showing a strong correlation between predicted cost of a Gator network and its actual cost when used for trigger condition testing. The results show that optimized Gator networks normally have a shape which i...

This paper describes the architecture of the TriggerMan asynchronous trigger processor. TriggerMan processes triggers after updates have committed in the source database. It is designed to be able to gather updates from a wide variety of sources, including relational databases, object-relational databases, legacy databases, flat files, the web, and others. TriggerMan achieves the ability to gather updates from so many sources using an extensible data source mechanism. TriggerMan supports single-table (single data-source) triggers, sophisticated multiple-table (multiple-data-source) triggers, and temporal triggers. It maintains persistent copies of derived data (including materialized views and time series objects) to enable highperformance trigger condition testing. In order to take advantage of an existing database storage manager rather then create a new one, TriggerMan is being implemented as an extension to an object-relational DBMS. To do parallel trigger condition testing, Trigge...

Current database trigger systems have extremely limited scalability. Some commercial systems allow only one trigger per table to be defined for each type of update event. Application systems could use the ability to create thousands or even millions of triggers. The advent of the Internet and the World Wide Web has made it even more important that trigger systems scale up since users could create large number of triggers via a web interface. This paper presents a way to develop a truly scalable trigger system. Scalability is achieved with a trigger cache to use main memory effectively, and a memory-conserving selection predicate index based on the use of unique expression formats called expression signatures. A key observation is that if a very large number of triggers are created, most will have the same structure, except for the appearance of different constant values. Expression signatures are used to divide selection predicates of triggers into equivalence classes. When a trigger i...

....................................................................................................................... vi CHAPTERS 1. INTRODUCTION ......................................................................................................... 1 2. OVERVIEW OF THE TRIGGERMAN ....................................................................... 6 2.1 Introduction ....................................................................................................... 6 2.2 TriggerMan Architecture................................................................................... 7 2.3 TriggerMan Command Language ..................................................................... 8 2.4 The Selection Predicate Indexing Scheme ........................................................ 8 3. RELATED WORK...................................................................................................... 12 3.1 Introduction to Ingres ...................................................

Web-based subscription services are becoming important as more businesses migrate to the Web. Monitoring subscriptions is to determine, for a given event, the subset of all subscriptions whose predicates match the event. Usually, subscriptions contain both equality and non-equality clauses. The general approach to this problem is to identify a "small" candidate set of subscriptions for each occurring event and then determine the matching subscriptions from the candidate set. Previous work has focused on using the equality predicate clauses to identify the candidate set. We found that completely ignoring non-equality clauses can result in a much larger candidate set. However, the effect of using non-equality predicate clauses varies in the sense that not every one of them can reduce the size of the candidate set and there is a diminishing marginal return if too many non-equality predicate clauses are included. We provide a method to map each subscription into a data point in a multiple dimensional space based on its predicate, including the non-equality clauses, and devise an adaptive multiple key hashing (AMKH) method to judiciously select an effective subset of the non-equality clauses to be included in identifying the candidate set. AMKH further provides a controlling mechanism to limit the hash range of a non-equality clause, hence reducing the size of the candidate set. Simulation studies show that AMKH performs well under various conditions.

Grid computing systems utilize distributively owned and geographically dispersed resources for providing a wide variety of services for various applications. One of the key considerations in Grid computing systems is resource management with quality of service constraints. The quality of service constraints dictate that submitted tasks should be completed by the Grid in a timely fashion while delivering at least a certain level of service for the duration of execution. Because the Grid is a highly "dynamic" system due to the arrival and departure of tasks and resources, it is necessary to perform advance reservations of resources to ensure their availability to meet the requirements of the different tasks. This paper introduces a new scheduling algorithm for advance reservations. Simulations are performed to compare our algorithm with an existing approach. The results indicate that the proposed algorithm can improve the overall the performance by satisfying larger number of reservation requests.

Nachdruck auch auszugsweise verboten c○2006

We introduce the dense skip tree, a novel cache-conscious randomized data structure. Algorithms for search, insertion, and deletion are presented, and they are shown to have expected cost O(log n). The dense skip tree obeys the same asymptotic properties as the skip list and the skip tree. A series of properties on the dense skip tree is proven, in order to show the probabilistic organization of data in a cache-conscious design. Performance benchmarks show the dense skip tree to outperform the skip list and the self-balancing binary search tree when the working set cannot be contained in cache. 1