|
|
 |
| |
|
|
|
Predictive Web Service Monitoring using Probabilistic Model Checking |
|
|
|
PP: 139-148 |
|
|
|
Author(s) |
|
|
|
Honghao Gao,
Huaikou Miao,
Hongwei Zeng,
|
|
|
|
Abstract |
|
|
| Web service is vulnerable to stochastic failures due to the changeful Internet environment, which will seriously impact the
reliability of business-critical applications. Web service must adapt to these changes. This paper studies the monitoring issue of Web
service and proposes a predictive service monitoring approach to support service-oriented software dynamic evolutions. It aims to
ensure that Web service is high-quality at runtime. Considering functional and non-functional requirements, we employ probabilistic
model checking technique to quantitatively verify the interactive behaviour of Web service. However, limited by time and memory,
frequently performing probabilistic model checking is not a high-efficiency solution to monitor Web service. Thus, forecasting the
reliability becomes an important way to enhance the performance of Web service monitoring, by which we can put forward dynamic
reconfigurations before the exception occurs. For this purpose, according to the historical probability values of Web service reliability
recorded in monitoring phase, we adopt the mathematical statistics, unary linear regression analysis method, to predict the probability
value of reliability based on invocation duration time. Finally, a case study is discussed and experiment results demonstrate the
applicability and effectiveness of our approach. |
|
|
|
|
|
|
|
