Eddies: continuously adaptive query processing
SIGMOD '00 Proceedings of the 2000 ACM SIGMOD international conference on Management of data
Models and issues in data stream systems
Proceedings of the twenty-first ACM SIGMOD-SIGACT-SIGART symposium on Principles of database systems
Convex Optimization
Monitoring streams: a new class of data management applications
VLDB '02 Proceedings of the 28th international conference on Very Large Data Bases
Scheduling intense applications most 'surprising' first
Proceedings of the 2010 ACM Symposium on Applied Computing
| Hi-index | 0.00 |
In a data stream management system (DSMS), there are a large number of simultaneously executing queries, each of which performs a specific monitoring function on some subset of the streaming data. In order to keep up with the rapid arrival rates, a DSMS may sample or drop tuples to match execution rate with the arrival rate. However, as the sampling rate reduces, so does the quality of answers of the queries. Since, there may be substantial overlap between the queries that are simultaneously active in the system, an optimal choice of the points and extent of load-shedding cannot be obtained without taking a "global view" of this problem. In this paper, we formulate the problem of optimizing the rates of sampling with the the overall objective of minimizing the sum of the loss of quality of answers of the queries, assumed to be associated with each query (e.g., as in the Aurora system). Specifically, we consider the problem of obtaining the optimal sampling probabilities for a given collection of sliding-window queries with known loss of quality plots, so that the objective function of the sum of loss of quality across the queries is minimized. The problem can be generally posed and solved as a convex optimization problem and can be solved in low order polynomial time.