Loading...
Please wait, while we are loading the content...
Similar Documents
Parallel XML Matching Algorithm for Publish / Subscribe Systems
| Content Provider | Semantic Scholar |
|---|---|
| Author | Ferreira, Susana |
| Copyright Year | 2015 |
| Abstract | The Publish/Subscribe communication model is the most adequate distributed paradigm for complex event processing and large-scale dissemination of information to a variety of users. The continuous growing interest in XML as the standard language for information representation and exchange over the internet increased the importance of XML-based publish/subscribe systems. The core functionality of Publish/Subscribe systems lies in the matching algorithm that is responsible for, whenever new events are published, determine matched subscriptions. In real world applications with a huge amount of stored subscriptions and continuously arriving events, the sequential matching algorithm can easily become a bottleneck impacting the overall performance of the system. A way to achieve a scalable system, while maintaining high performance, is by exploiting chip multiprocessors architectures already present in today’s computers. In this paper, we propose and implement three parallel event processing techniques for DeltaFilter, a highly efficient sequential XML matching algorithm. We perform experimental evaluations on a 48 core machine to study the scalability and performance of the proposed techniques with a varying number of threads in different application scenarios. The results show performance gains of 20 times more events processed per second and a reduction of almost 74% on the matching time per event when in the presence of 48 threads. |
| File Format | PDF HTM / HTML |
| Alternate Webpage(s) | https://fenix.tecnico.ulisboa.pt/downloadFile/563345090413950/resumo.pdf |
| Language | English |
| Access Restriction | Open |
| Content Type | Text |
| Resource Type | Article |
