To Share, or not to Share Online Event Trend Aggregation Over Bursty Event Streams Contributions

We present a novel framework Hamlet for optimizing a workload of queries computing aggregation over Kleene pattern matches, called event trends. Hamlet is the first to seamlessly integrate the power of online event trend aggregation and adaptive sharing decision-making.

To tackle the exponential complexity of event trend construction, Hamlet deploys online trend aggregation strategy. To this end, we design the Hamlet graph that compactly captures all trends matched by queries in the workload and propagates trend aggregates through this graph without constructing the actual trends. This graph-based online execution strategy allows to reduce the time complexity of event trend aggregation from exponential to quadratic in the number of matched events.

To quantify the trade-off between the benefit of sharing and the overhead of maintaining the intermediate trend aggregates per query, we design a light-weight sharing benefit model.

To enable shared query execution over bursty high-rate event streams, we propose an effective sharing decision algorithm that adaptively at runtime activates the shared online execution based on the benefit of sharing the trend aggregation.

Our experiments on real and synthetic event streams show that Hamlet achieves up to five orders of magnitude performance improvement over state-of-the-art approaches

---