You are on page 1of 5

A Framework for Optimizing the Availability of Composite Web Service

N.Sasikaladevi Dr.L.Arockiam
Lecturer, Associate Professor,
Department of Computer Applications, Department of Computer Science,
Bishop Heber College, St,Joseph’s College,
sasikalade@yahoo.com larockiam@yahoo.co.in

Abstract service software module and its description. In a


typical scenario a service provider hosts a network
Web services are playing an important role in accessible software module (an implementation of a
today’s business environment. Business tasks are web service). The service provider defines a service
computerized as jobs. Jobs are implemented as description for the web service and publishes it to a
composite web services in which more than one requestor or service discovery agency. The service
specific service are involved. Numerous web services requestor uses a find operation to retrieve the service
are evolving now. In which choosing the appropriate description locally or from the discovery agency and
web service is a tedious task today. Consumer may uses the service description to bind with the service
require service with some quality parameter. Service provider and invoke or interact with the web service
Level Agreement plays a vital role in today’s implementation. Service provider and service
industry to get the desired services with the user requestor roles are logical constructs and a service
given parameters. Security, Reliability and may exhibit characteristics of both.
Availability are the important Quality of Service • Service requester -- requests the execution
metrics in which Availability plays a dominant role, of a Web service
because the desired web service should be made
available to the user first. In this paper a framework • Service provider -- processes a Web service
model is designed to optimize the availability metric request
of composite web service.
• Discovery agency -- agency through which a
Web service description is published and
1. Introduction made discoverable

A Web Service is a software interface that A software agent in the Web services
describes a collection of operations that can be architecture can act in one or multiple roles, acting as
accessed over the network through standardized requester or provider only, both requester and
XML messaging. It uses protocols based on the XML provider, or as requester, provider, and discovery
language to describe an operation to execute or data agency. A service is invoked after the description is
to exchange with another Web service. The basic found, since the service description is required to
Web services architecture defines an interaction establish a binding. Service requestor and service
between software agents as an exchange of messages provider interact, based on the service's description
between service requesters and service providers. information published by the provider and
Requesters are software agents that request the discovered by the requester through some form of
execution of a service. Providers are software agents discovery agency.
that provide a service. Agents can be both service A web service is characterized by its flexibility
requesters and providers. Providers are responsible to encapsulate discrete business functionalities and
for publishing a description of the service(s) they its interoperability to support universal application
provide. Requesters must be able to find the integration. Composition of computational resources
description(s) of the services. The basic Web service and web-based services into integrated solutions is a
architecture models the interactions between three key activity to enhance offerings and allow for a
roles: the service provider, service discovery agency, smooth process from the point of view of customers.
and service requestor. The interactions involve the Such integration process has been greatly simplified
publish , find, and bind operations. These roles and with the advent of web services technology.
operations act upon the web service artifacts: the web
SLA plays a vital role in today’s e-business provider and service user. Behavior security is the
environment. SLA is an agreement between the trustworthiness of the result obtained by the web
service provider and service consumer. A service service and trustworthiness of the resource sites
level agreement is an agreement regarding the which are involved in the specific web service.
guarantees of a web service. It defines mutual Performance includes response time and throughput.
understandings and expectations of a service between Response time is the time for a service request to be
the service provider and service consumers. The satisfied. Throughput is the servicing rate of the
service guarantees are about what transactions need service provider. At last, Availability is the
to be executed and how well they should be percentage of the time that a service provider can
executed. An SLA may have the following offer services.
components:
§ Purpose - describing the reasons behind the Availability is an important metric in QoS. A
creation of the SLA service becomes unavailable to the service consumer
§ Parties - describes the parties involved in due to various reasons such as network failure,
the SLA and their respective roles (provider hardware failure, software failure or operational
and consumer). failure. Web services are implemented in distributed
§ Validity period - defines the period of time environment. Service request and service response
that the SLA will cover. This is delimited by should have traverse various intermediate sites which
start time and end time of the term. are connected through network. If any one of the
§ Scope - defines the services covered in the network link fails, then the service becomes
agreement. unavailable to the service user. Service request
§ Restrictions - defines the necessary steps to involves the use of various hardware resources in
be taken in order for the requested service servicing site. The failure of hardware makes service
levels to be provided. unavailable to the consumer. The service designed by
§ Service-level objectives - the levels of the service provider may fail because of software
service that both the users and the service bugs. Service may be unavailable due to server
providers agree on, and usually include a set failure.
of service level indicators, like availability, In this paper, we considered the availabilityof
performance and reliability. Each aspect of the server to evaluate the availability of the service.
the service level, such as availability, will Further more, numerous similar services are available
have a target level to achieve. for same task. If any one the service is unavailable,
§ Penalties - spells out what happens in case we can choose the proxy service which does the same
the service provider under-performs and is thing. In this paper, we considered the proxy service
unable to meet the objectives in the SLA. If also, to achieve the better availability result. A new
the agreement is with an external service framework model is designed for optimizing the
provider, the option of terminating the availability of the composite web service.
contract in light of unacceptable service
levels should be built in.
§ Optional services - provides for any services 2. Related Work
that are not normally required by the user,
but might be required as an exception. Subil Abraham et all described a method to
§ Exclusions - specifies what is not covered in enhance web service availability. They proposed
the SLA. architecture for highly available web services for
§ Administration - describes the processes mission critical applications. The central idea is the
created in the SLA to meet and measure its enhancement of web services by the introduction of a
objectives and defines organizational central hub to increase the availability of web
responsibility for overseeing each of those services.[1]
processes. Cotroneo et all have given a model to improve the
availability of the web service. They presented the
Qualities of Services are defined in the service
overall architecture and the evaluation of a
level objectives part of SLA. SLA guarantee a level
middleware infrastructure which provides quality-of-
of service delivered to a customer for a given price.
service differentiation among classes of
Service providers define all the aspects of the service
communication-bound processes. By
being provided. Generally QoS for web services are
communication-bound processes are the processes
security, performance and availability. Security can
whose activity is typically dominated by network
be implemented by identity security and behavior
communication, e.g. a video server. The proposed
security. Identity security is a combination of
architecture supports different classes of service,
authentication and authorization between the service
each with different quality attributes concerning the
network data delivery performance. In particular, the N i = {S i , Bij }(i ∈ [1..n ], j ∈ [0..q ]
architecture is able to provide a class of service,
namely guaranteed service class, which is suitable for S i is the primary service in the composite web
increasing the service availability for a group of service. Each node could only contain one primary
premium users, especially in overloaded servers (in service. When composite service process to a
absence of external faults).[2] particular node, it will always call the primary
QoS issues are listed out by Menasce[3]. The service first. If the primary service fails, backup
issues are, services substitute the primary service. Bij is the
• Availability is the percentage of time that a
service is operating. back up service for S ; and q is the number of
• Security properties include the existence backup service.
and type of authentication mechanisms the If q = 0 then no backup service available. If
service offers, confidentiality and data
integrity of messages exchanged, non
q ≥ 1 then back up services are available.
repudiation of requests or messages, and Structuring the composite web service is a crucial
resilience to denial-of service attacks. task. To implement the operation relationship of a
• Response time is the time a service takes to business task, we used Jorge Cardoso’s stochastic
respond to various types of requests. workflow reduction (SWR) algorithm [6]. Figure 1
Response time is a function of load illustrates four relationships; they are sequential,
intensity, which can be measured in terms of parallel, conditional, simple loop and dual loop. Each
arrival rates (such as requests per second) or node is a data assignment, exchanging an event,
number of concurrent requests. QoS takes doing an action, or executing a sub-process, applying
into account not only the average response those relationship we can compute total availability
time, but also the percentile (95th of a composite web service.
percentile, for example) of the response Composite service availability could be defined
time. as,
• Throughput is the rate at which a service {
Ac = AN i , Os }
can process requests. QoS measures can
Ac denotes the aggregate availability,
include the maximum throughput or a
function that describes how throughput AN i denotes the availability of each node and Os is
varies with load intensity [3]. the operation relationship of sub set of node set.
Availability evaluation based on the Cardoso
Magnos Martinello_and all has given the operational relationships are given below,
impacts of error recovery in web service availability.
In their work, a composite performance and
• Sequential:
availability modeling approach is considered where
various causes of service unavailability and two A = A(ti ) A(t j )
different recovery strategies are taken into account. • Parallel:
For illustration purposes, sensitivity analysis results A = A(t a ) A(tb )∏ A(ti )
are presented to show the impact on the web service i
availability of various assumptions concerning the
• Conditional:
architecture, the servers load, as well as the failure
and recovery processes [4]. A = A(t a ) A(tb )∑ pi A(ti )
i

3. Composite Web Service • Simple loop:


Representation (1 − pi ) A(ti )
A=
1 − pi A(ti )
Business tasks are implemented as composite web n
service. Composite web service is defined as, Where, pi + ∑ poj = 1
C = {N i }(i ∈ [1..n ] j =1

C defines the composite web service for • Dual loop:


availability evaluation. The basic unit of composite (1 − pi ) A(ti )
A=
web service is defined as N i , n is the number of 1 − pi A(ti ) A(t j )
nodes in the composite web service. Service node is n
defined as, Where, pi + ∑ poj = 1
j =1
Two states Markov model is used to represents up
and down state of the server. The failure rate ai is
the state of transition from up to down and recovery
rate bi represents the rate of transition from down to
up. The probability p i j that i servers of a total
N j servers are down is given by,
i Nj! i
p j= η j p 0 j , for i = 1,…. Nj
i!( Nj − i )i

aj
Where, η j= is the server unavailability rate
a j + bj
and p 0 j is given by,
−1
 Nj
ηi j 
p j =  Nj!∑
0


 i =0 i!( Nj − i )!
The probability that no more than N j − n j servers at
site j are down is:
Nj − nj
Nj!
Pj (n j , Nj ) = p 0 j ∑ i!( Nj − i)!η
i =0
i
j --1

This is the probability that at least n j servers in site


j of a total N j servers are available.

5. Framework Model for Optimizing the


Figure 1: Operational relationships of composite web service Availability of Composite Web Service
Availability is a predominant component QoS. A
4. Availability of Composite Web Service
service becomes unavailable to the service consumer
due to various reasons such as network failure,
Availability of the web service mainly depends on
hardware failure, software failure or operational
the availability of the server which offers the service.
failure. Service may be unavailable due to server
If the server is in down state, then the service will be
failure. In this paper, a framework model is given to
unavailable to the user. The service availability is the
handle the availability of server.
percentage of time that a server is up or down. Let
MMTF (Mean time to failure) be the average time Several components are involved in the figure 2.
of a server failure, and MMTR (Mean time to The components are.
recover) be the average time for recovering a server. • Service Providers: Service providers are
the software agent providing services to
1 the customer with quality parameter
Failure rate of the server is , which is
MMTF mentioned in Service Level Agreement.
denoted by a . Recovery rate of the server • Service Users: Service users are
1 consumers who are using the service to
is , which us denoted by b . Composite meet their need.
MMTR • Service Broker: Service Broker is a
service involves several services whose relationships middleware between the service
are given in the figure. Because of the enormous consumer and service producers.
growth of the web service, same task may be
• Service Orchestrator: Service
performed by more than one service. So, Composite
Orchestrator initiates the execution of a
service involves several services, for each and every
sequence of tasks associate with the
service in composite service may have several
workflow of a service request.
alternates or proxies.
Figure 2: Framework model for optimizing the availability of composite web service
• Composite service Manager: Composite
service manager spilt the composite 6. Conclusion
service into a set of service based on the
Cardoso operational model. A simple framework model is proposed in this
• Availability Manager: Availability paper to optimize the availability of the composite
manager evaluates the availability of the web service. In future, the model has to be tested
service. with real time example with the use of simulator tool.
Steps involved in this model is, For this model, only the server availability is
1. A customer submits a service request to the considered to evaluate service availability. But
optimizing agent. service availability is based on the availability of
2. Composite service manager, who is part of network, hardware, software and operational aspect
the optimizing agent, split composite service of the service. In future, all these components have to
into a set of services. be considered to optimize the availability of the
3. Availability manager contact service broker composite web service.
through which servicing servers for a specific
web services are identified. It evaluates the 7. References
availability of each server using two state
Markov model as given in the equation 1. [1] Subil Abraham, Mathews Thomas, Johnson Thomas,
The availability value of the service is given “Enhancing Web Services Availability”, Proceedings
to the composite service manger. of the IEEE International conference on e-Business
4. Composite service manager aggregate the Engineering, IEEE computer society, 2005.
availability values. If the probability of the
availability of the services is greater than or [2] D.Cortroneo, M.Gargiulo, S.Russo, G.Ventre,
“Improving the availability of web services”,
equal to the desired percentage of service
Research article, Italian Ministry of research, 2004.
availability, then it sends the positive
response to the customer otherwise it sends [3] Daniel A.Menasce, “QoS Issues in Web Services”,
the negative response to the customer. IEEE Internet Computing, IEEE, December 2002, pp.
5. If the response from the composite service 72-75.
manager is positive then the customer
initiates a job request to the service [4] Magnos Martinello, Mohamed Kaaniche and Karama
orchestrator. Service orchestrator selects the Kanoun, “Web Service Availability-Impact of Error
servers to complete the task. The response Recovery”, Springer, 2003, pp.165-178.
from servers are consolidated, and then given
[5] Judith Hurwitz, Robin Bloor, Carol Baroudi and
back to customer. If the response from the Marcia Kaufman, Service Oriented Architecture for
composite service manager is negative then Dummies, Wiley Publisher, 2007.
the customer could not initiate the job
request. [6] J.Cardosa, J.Miller, A.Sheth and J.Arnold, “Modeling
If the probability of the availability of the Quality of Service for Workflows and Web Service
primary service is less than the percentage of Processes”, Technical Report, University of Georgia,
desired probability, then the availability manger 2002.
choose the backup service from the pool of similar [7] Daniel A.Menasce, “Composing Web Services-QoS
View”, IEEE Internet Computing, IEEE, December
service and it will evaluate the availability
2004, pp. 88-90.
probability and it is given to the composite service
manager.

You might also like