Reactive object-oriented databases and CIM

Performance evaluation of object-oriented active database systems using the BEAST benchmark

“…ACOOD is based on an array technique [15] for detection of composite events which is efficient for the detectable events. Rules in ACOOD are indexed by events [2], which speeds up the process of selecting triggered rules. Due to the architectural style of ACOOD (layered approach), we have not been able to efficiently implement transaction events.…”

“…• ACOOD [2,15] is built on top of the commercial OODBMS ONTOS DB 3.0, • NAOS [11,10] which extends the OODBMS O 2 [1] (version 4.5.6), • Ode [29] exists in two variants: a disk-based version built on EOS [4] and a main-memory version built on Dali [26] (the disk-based one was benchmarked), and • SAMOS [18] uses the commercial OODBMS ObjectStore as a platform.…”

“…After an event is detected, the corresponding rule will be fired. Events can be either primitive (e.g., a method invocation, a transaction begin or commit, a time event, an abstract event 2 ) or composite (e.g., conjunction, sequence, disjunction, negation, repeated occurrence). Composite event restrictions are conditions that must hold for component events in order to form a legitimate composition.…”

“…In this paper we therefore first identify performance-critical aspects of ADBMS and then introduce the BEAST benchmark [22] (BEnchmark for Active database SysTems). We also report on results obtained from applying BEAST to four object-oriented ADBMS (ACOOD [2], NAOS [11], Ode [29], and SAMOS [18]). 1 We interpret the benchmark results obtained for each system and make some general conclusions.…”

The Starburst active database rule system

“…Then, when the transaction is ready to commit, rule execution is invoked automatically. 7 An important advantage of this approach o ver the straightforward approach o f i n voking the Rule Execution Module at the end of every transaction is that no overhead is incurred by transactions for which no rules are triggered.…”