Schema-based Scheduling of Event Processors and Buffer Minimization for Queries on Structured Data Streams

Abstract

We introduce an extension of the XQuery language, FluX, that supports event-based query processing and the conscious handling of main memory buffers. Purely event-based queries of this language can be executed on streaming XML data in a very direct way. We then develop an algorithm that allows to efficiently rewrite XQueries into the event-based FluX language. This algorithm uses order constraints from a DTD to schedule event handlers and to thus minimize the amount of buffering required for evaluating a query. We discuss the various technical aspects of query optimization and query evaluation within our framework. This is complemented with an experimental evaluation of our approach.

Parallelizing XML data-streaming workflows via MapReduce

“…Our work also has a number of similarities to the area of query processing over XML streams (e.g., see [36][37][38][39][40][41][42]). Most of these approaches consider optimizations for specific XML query languages or language fragments, sometimes taking into account additional aspects of streaming data.…”

Cost-based vectorization of instance-based integration processes

“…Examples for those systems are QStream [31], Demaq [32] and Borealis [33]. Typically, those systems use a one-thread-per-operator model (which is comparable to full vectorization) or a single-threaded operator scheduling with central control strategies (assuming high costs for switching the process context) [34][35][36]. There are interesting approaches [17,[37][38][39], where operators are distributed across a number of threads in a query-aware manner.…”

Automata for XML—A survey

“…In [73] queries are specified in terms of attributed grammars and evaluated by deterministic depth-synchronous pushdown document automata. A stream-oriented language close to XQuery, FluX, is defined in [74]. It is shown how evaluation can be optimized in the presence of schema knowledge.…”