基于量化模型的Web服务故障自适应重配置策略

87 浏览量更新于2024-08-26 收藏 308KB PDF 举报

在现代信息技术领域，面向服务的软件（Service-Oriented Software, SOS）已经成为企业构建和管理复杂业务流程的重要工具。Web服务重新配置在SOS系统中扮演着核心角色，特别是在不确定性极高的互联网环境中，它确保关键业务应用能够持续稳定运行。本文主要关注如何通过量化模型来优化Web服务的重新配置选择。首先，文章将Web服务的服务过程建模为行为级别和QoS（Quality of Service，服务质量）双重维度上的概率定时模型PTWSB。这种模型考虑了服务的执行行为以及其在满足预定义服务质量标准下的可能性，为服务选择提供了量化评估依据。行为级模型关注服务的执行流程和依赖性，而QoS模型则着重于响应时间、可靠性、吞吐量等性能指标。接下来，作者提出了两种基于模型的重配置策略：单源服务重配置和多源服务重配置。单源服务重配置针对的是当某个服务发生故障时，系统如何从一组候选服务中挑选一个最佳的替换服务，以最小化对业务的影响。这种方法通常适用于资源有限、服务依赖线性的场景。相比之下，多源服务重配置更为复杂，它考虑了多个可能的服务来源和服务间的冗余和备份。这种策略旨在提高系统的可用性和恢复能力，即使在多个服务同时失效的情况下，也能通过多路径寻找解决方案，从而增强系统的整体稳定性。作者的研究方法通过量化分析和模型预测，不仅提高了Web服务选择的精确度，而且有助于在动态变化的互联网环境中快速做出决策，降低服务中断的风险。由于其高效性和适用性，这项工作在实际的SOS系统中具有广阔的应用前景，可以促进企业的服务架构更加灵活和适应性强。本文的核心贡献在于提出了一种量化模型驱动的Web服务重新配置策略，该方法对于提升SOS系统的鲁棒性、可靠性和可扩展性具有重要意义，为业界提供了一种有效的工具来应对不断变化的业务需求和互联网环境中的挑战。

A Quantitative Model-based Selection of Web Service Reconfiguration

Honghao Gao

, Huaikou Miao

Computer Center, Shanghai University, 200444 Shanghai, P.R. China

School of Computer Engineering and Science, Shanghai University, 200444 Shanghai, P.R. China

E-mail: gaohonghao@shu.edu.cn

Abstract — Web service reconfiguration plays a critical role in

Service-Oriented Software (SOS), which provides a self-

adaptation technique to ensure the business-critical

application can be correctly worked when an SOS system is

deployed in the uncertain Internet environment. To address

this problem, the primary task is to select substitution services

for handling the current failure service, such as the atomic

service or composite service. In this paper, the service process

of Web service is initially formalized in the form of a

probabilistic timed model PTWSB at both of the behavior

level and QoS level. Then it gives two model-based

reconfigurations with different service selection demands,

mainly the single-source service reconfiguration and multi-

source service reconfiguration. Our approach has a good

potential application prospect in Service-Oriented Software.

Keywords-service selection; quantitative model checking;

synchronous product; task-oriented solution strategy

I. INTRODUCTION

Web service is considered to be the most promising

technique for Service-Oriented Software (SOS) in Service-

Oriented Computing (SOC) and Service-Oriented

Architecture (SOA), which works as plugin mode to

provide the value-added applications in which required

services can be flexibly used on demand. Since the single

service has limited functionality, complex business logic

calls for composing and integrating different kinds of cross-

domain services, utilizing not only atomic service but also

composite service. Thus, services composition approaches,

such as Orchestration and Choreography languages [1] (e.g.,

BPEL4WS, WS-Choreography, and WSCL), have been

used in commercial applications and open source research.

It contributes to business and software development. For e-

commerce, the service-oriented solution provides processes

to maximize business benefits that companies can easily

employ the existing third-party services and their interfaces

to realize inter-organizational cooperation and collaboration.

For software, the service-based architecture supporting a

rapid reuse mode and low-cost way to software

implementation has completely changed the traditional

software life cycle. However, running under the

heterogeneous, open, and collaborative Internet

environment, Web services may become faulty occasionally.

SOS is required to respond to the changes in order to avoid

service interruptions, including varying user requirements

and unstable Internet runtime environments. If the service

failure is not processed in a timely manner, it may seriously

impact the correctness and reliability of business processes

of a service-oriented software. Thus, effectively executing a

reconfiguration ensures the business-critical application is

trustworthy. One of the most challenging problems is to

select candidate Web services [6-15, 17-23].

The first issue is how to guarantee core functions will

work after deploying the selected service in service

reconfiguration. This refers to the behavior matching

requirement. The service process of business logic may

have complex dependencies. Thus, the functional

requirement verification of a Web service needs to be

executed by the temporal logic for checking the behavior

consistency. The second issue is how to ensure the

candidate service has equivalent (or similar) non-

functionality to the failure one. This refers to the

quantitative requirement. From the respect of consumers,

satisfaction relies on their user experiences and on judging

the quality of service when invoking a service, such as

timing-response, cost and reliability. The real-time property

of a Web service depends on input/output and the timing of

input/output [2]. The cost property of a Web service refers

to utilize Web service at little expense. The probability

property of a Web service provides a robustness mechanism

when users invoke service-based software. In short, a

response should provide correct values, and the value

should be output at the right time-points with lower cost

and high reliability.

The process of manually selecting a composite Web

service or replacement for the failure one is time consuming

and error-prone. Thus, in this paper, a novel model-based

approach is proposed for quantitatively selecting Web

services during service reconfiguration. It aims to employ

automata to select a suitable candidate set of services. The

service process of Web service is formalized into a

probabilistic timed

model (PTWSB) and the quantitative

reconfiguration requirement is specified in the form of

probabilistic timed computation tree logic (PTCTL). The

quantitative reconfiguration based on model checking is

divided into two patterns when considering service

selection, mainly the single-source service reconfiguration

method and multi-source service reconfiguration method.

The former tries to select a single service to replace the

failure service. If no suitable service exists, the latter

attempts to select multiple services to simulate the failure

service. Considering linear sequences illustrating property

violations can be produced by the model checker, the

quantitative counterexample is introduced to describe the

possible strategy for selecting and recommending services

when using the multi-source service reconfiguration method.

The remainder of this paper is organized as follows.

Section II introduces preliminary concepts. Section III

shows the ideal modeling service process and specifying

DOI 10.1109/SNPD.2013.19

365

2013 14th ACIS International Conference on Software Engineering, Artificial Intelligence, Networking and Parallel/Distributed

Computing

DOI 10.1109/SNPD.2013.19

365

下载后可阅读完整内容，剩余6页未读，立即下载

weixin_38713203

粉丝: 11
资源: 942

基于量化模型的Web服务故障自适应重配置策略

预测性Web服务监控：概率模型检查方法

基于Web的企业内部人员绩效量化系统实现与源码解析

JSP与PHP动态Web服务器性能分析及Simulink建模研究

面向Web服务的动态可信性评估模型.pdf

基于Matlab的事件驱动量化回测框架.zip

基于ssm+Vue的量化积分管理系统.zip

java_ssm基于Vue的量化积分管理系统vue毕业论文.doc

基于PHP的币商智能量化机器人交易系统 v2.0.4.zip

基于SpringBoot的企业绩效量化系统Java代码实现

基于SpringBoot的企业绩效量化管理系统源码发布

最新资源