Professional Documents
Culture Documents
QoS-based Discovery and Ranking of Web Services
QoS-based Discovery and Ranking of Web Services
Abstract— Discovering Web services using keyword-based search share similar functionalities is significantly achieved by
techniques offered by existing UDDI APIs (i.e. Inquiry API) may examining non-functional Web service attributes such as
not yield results that are tailored to clients' needs. When response time, throughput, availability, usability, performance,
discovering Web services, clients look for those that meet their integrity, among others. It would be desirable if existing
requirements, primarily the overall functionality and Quality of standards applied for publishing, discovering, or using Web
Service (QoS). Standards such as UDDI, WSDL, and SOAP have services have the ability to incorporate QoS parameters as part
the potential of providing QoS-aware discovery, however, there of the registration process (i.e. publishing process) while
are technical challenges associated with existing standards such continuously regulating or monitoring revisions to QoS
as the client's ability to control and manage discovery of Web information as a result of any related Web service updates.
services across accessible service registries. This paper proposes a
solution to this problem and introduces the Web Service There are several approaches that address how to deal with
Relevancy Function (WsRF) used for measuring the relevancy QoS for Web services. However, many of them rely on the
ranking of a particular Web service based on client’s preferences, service providers to supply their QoS metrics and therefore,
and QoS metrics. We present experimental validation, results, storing this type of information either in the UDDI or at the
and analysis of the presented ideas. service provider’s site may not be the best solution due to the
fact that features such as response time and throughput will be
Keywords-UDDI, Service Registries, Web Services, Quality of advertised by the service provider and may be subjected to
Service, QoS, Web Service Broker, Discovery of Web Services,
forms of manipulation. In addition, the supply of QoS metrics
Ranking of Web Services, Ranking
by the service provider raises several concerns such as integrity
and reliability of the supplied values. It would be ideal if there
I. INTRODUCTION is a trusted service broker that can manage the supply of QoS
Standards such as UDDI have enabled service providers information for Web services in a transparent manner such that:
and requestors to publish and find Web services of interest (1) service providers provide only QoS information that must
through UDDI Business Registries (UBRs), respectively. directly be supplied (i.e. cost per invocation or price plans)
However, UBRs may not be adequate enough for enabling through an interface; and (2) QoS metrics that are not
clients to search for relevant Web services due to a variety of necessarily supplied by service providers (i.e. response time,
reasons. One of the main reasons hindering the efficient availability, reliability, penalty rate, among others) are
discovery of Web services is the fact that existing search APIs computed in an autonomous manner.
(i.e. UDDI Inquiry API) only exploit keyword-based search To address the above issues, this paper introduces a
techniques which may not be suitable for Web services mechanism that extends our Web Services Repository Builder
particularly when differentiating between those that share (WSRB) architecture [1] by offering a quality-driven discovery
similar functionalities. of Web services and uses a combination of Web service
Furthermore, many software vendors are promoting their attributes as constraints when searching for relevant Web
products with features that enable businesses and organizations services. Our solution has been tested and results show high
to create their own UBRs (i.e. IBM WebSphere, Microsoft success rates of having the correct or most relevant Web
Enterprise Windows Server 2003, and others). In this case, service of interest within top results. Results also demonstrate
businesses and organizations may preferably deploy their own the effectiveness of using QoS attributes as constraints when
internal UBRs for intranet or extranet use which will cause a performing search requests and as elements when outputting
significant increase in the number of discrete UBRs over the results. Incorporating QoS properties when finding Web
Web. This adds to the already existing complexity of finding services of interest provides adequate information to service
relevant Web services of interest in the sense that there needs requestors about service guarantees and gives them some
to exist an automated mechanism that can explore all accessible confidence as to the quality of Web services they are about to
UBRs, mainly a Web services’ discovery engine. invoke.
Due to the fact that much of the information provided by The rest of this paper is organized as follows. Section II
Web services is mostly technical, there is a need for a discusses the related work. Our Web service ranking
mechanism that can distinguish between Web services using mechanism is presented in Section III. Experiments and results
well-defined criteria such as considering Quality of Service are discussed in Section IV, and finally the conclusion and
(QoS) attributes. Differentiating between Web services that future work are discussed in Section V.
C (cents/
RT (ms)
AV (%)
AC (%)
Service
invoke)
IA (%)
I keyword-based search) is Web service number seven. Figure 2
TP
Provider &
D shows the results from computing WsRF values for all Web
Name
services listed in Table I using a keyword-based search
XMLLogic
1 ValidateEmail
720 6.00 85 87 80 1.2 technique versus a client-controlled search in which cost
XWebservices represents the most important QoS parameter (i.e. running
2 XWebEmail- 1100 1.74 81 79 100 1 WsRF that is heavily dependent on cost).
Validation
StrikeIron
Client Controlled vs. Keyword-based
3 Email 710 12.00 98 96 100 1
Verification QoS Specified Generic Search (No QoS)
StrikeIron
Email Address 912 10.00 96 94 100 7 5
4
WsRF (ws i )
Validator 4
CDYNE 3
5 910 11.00 90 91 70 2 2
Email Verifier
Webservicex 1
6 1232 4.00 87 83 90 0 0
ValidateEmail
ServiceObjects 0 1 2 3 4 5 6 7
7 DOTS Email 391 9.00 99 99 90 5 Web Service Number
Validation
Table I shows QoS values that were measured by WS- Figure 2. Results from running WsRF heavily dependenton cost vs.
QoSMan for seven Web services. In order to find the most keyword-based search
relevant Web service, there needs to exist a set of Web services
that share the same functionality and are used to measure QoS Based on QoS values obtained in Table I, Web service
metrics based on QoS criteria. Once WSRB has successfully number six has the least cost (zero implies that it is being
generated the necessary QoS metrics, these QoS values are offered at no cost) which complies with the results obtained
used as inputs to the WsRF and the matrix in (1) is established. from running WsRF that is heavily dependent on cost as Figure
It is important to note that WSRB does not necessarily have to 1 demonstrates (i.e. Web service number six has the highest
perform QoS metrics check when a client performs a request, WsRF value). When analyzing results from Figure 2, it is
but rather uses an update interval for measuring these metrics. important to take into consideration WsRF values when
This provides the WSRB with up-to-date QoS information that associating at least one QoS parameter which results in WsRF
is ready and available upon client requests in real-time values ranging from 0 to 1 while having a broad search that is
scenarios. not QoS specific (i.e. WsRF without weights) produces WsRF
values ranging from 3.22 to 4.67. Having smaller values means
The values shown in Table I represent QoS values that the standard deviation is smaller, and therefore, the faster
measured for seven different Web services. In order to find the the WsRF is converging into a solution. Having smaller
most suitable Web service, it is important to optimize the standard deviation using WsRF outperforms keyword-based
values for each QoS parameter. For instance, having higher search technique in the sense that it provides a very good
probability for accessibility percentage is preferable than estimate of the true or optimal value, while providing precise
having a Web service with low probability for accessibility. In and accurate results.
this case, WsRF will maximize accessibility. However, for
some other QoS parameters such as cost, WsRF will minimize To demonstrate the effectiveness of our ranking technique
them. and how it outperforms other discovery methods that merely
depend on keyword-based technique, we will consider six test
WsRF (wsi )
A. Test Scenario 1 0.80
Figure 3 shows another test from running WsRF with 0.60
having response time as the most important QoS parameter. 0.40
Results from Figure 3 demonstrate that Web service number 0.20
seven has the highest WsRF value, or the one that has the 0.00
0 1 2 3 4 5 6 7
fastest response time, or 0.391 seconds, which conforms to the
data obtained in Table I. Web Service Number
QoS Ranking Heavily Dependent on RT Figure 5. QoS ranking with more emphasis on RT (75%) than TP (25%)
1.20
1.00
WsRF (wsi )
Figure 4 shows another test when running WsRF with more 0.80
emphasis on the maximum throughput (TP). Results from 0.60 0.4511 0.4606 0.4539
Figure 4 demonstrate that Web service number three has the 0.40 0.3044 0.3458 0.3537
highest WsRF value which is consistent with the data obtained 0.20
in Table I in which this Web service has the highest throughput 0.00
or 12 requests per minute. 0 1 2 3 4 5 6 7
Web Service Number
QoS Ranking Heavily Dependent on TP
1.20
1.00 Figure 6. QoS ranking dependent on RT (80%) and C (20%)
WsRF (wsi )
0.80
0.60 Results shown in Figure 6 can be compared to those in Test
0.40 Scenario 1 in the sense that in both of these scenarios, response
0.20 time is the dominant QoS parameter. However, WsRF values
0.00
slightly change in Test Scenario 4 such that response time
0 1 2 3 4 5 6 7
remains the dominant QoS parameter, but slightly takes into
Web Service Number
consideration the cost parameter. Table III demonstrates the
ranks for both scenarios and the ranking variation for each Web
Figure 4. QoS ranking heavily dependent on TP service.
C. Test Scenario 3 TABLE III. RANKING DEVIATION FOR TEST SCENARIOS 1 AND 4
0.7684 0.7783
0.80 0.6541
World Wide Web Conference, 2004.
0.6106
0.60 0.7113
[8] Zeng, L., Benatallah, B., Dumas, M., Kalagnanam, J., and Sheng, Q.Z.,
0.6993
“Quality Driven Web Services Composition”, Proceedings of the 12th
0.40 0.5361
International World Wide Web Conference, 2003, pp. 411-421.
0.20
[9] Sheth, A., Cardoso, J., Miller, J., and Koch, K., “Web Services and Grid
0.00
Computing”, Proceedings of the Conference on Systemics, Cybernetics
0 1 2 3 4 5 6 7 and Informatics, Florida, 2002.
Web Service Number [10] Sreenath, R., and Singh, M.P., “Agent-based Service Selection”. Journal
of Web Semantics, Volume 1, Issue 3, 2004.
Figure 7. Equal distribution of weights (w=0.1667) [11] Larkey, L., “Automatic Essay Grading Using Text Classification
Techniques”, ACM SIGIR, 1998.
Results in Figure 7 demonstrate a small standard deviation [12] Mansace, D., “QoS Issues in Web Services”, IEEE Internet Computing,
6(6), 2002, pp. 72-75.
when compared to the ones shown in Figure 2 using a generic
[13] Mindreef SOAPScope, http://home.mindreef.com, (Last Accessed May
search. For instance, the standard deviation when considering 2007).
WsRF without any weights is 0.5197 while considering equally
[14] Al-Masri, E., and Mahmoud, Q. H., “Crawling Multiple UDDI Business
distributed weights for all QoS parameters is reduced Registries”, 16th International World Wide Web Conference, 2007, pp.
significantly to 0.0866. This shows that associating weights 1255-1256.
with WsRF significantly improves the accuracy of the ranking [15] Al-Masri, E., and Mahmoud, Q. H., “Discovering the Best Web
and enables WSRB to converge into a solution. Therefore, the Service”, 16th International World Wide Web Conference, 2007, pp.
more client preferences specified, the narrower the results and 1257-1258.
the higher performance of WsRF.