at the client-server sides and the communication agents [6]. .... These CPNs are called pages (or subpages) and are related ..... dot, in dot's User's Manual. 2006.
2012
International Conference on Computer & Information Science (ICCIS)
Web Services Extension Model Simulation In Hierarchical Colored Petri Net Dani Adhipta, Mohd Fadzil Hassan, Ahmad Kamil Mahmood Computer and Information Sciences Department, Universiti Teknologi PETRONAS Bandar Seri Iskandar, 31750 Perak, Malaysia
Abstract-This paper presents the modeling in Web Services (WS)
extension
performance
by
(WS-E)
for
higher
Representational
Net
(HCPN)
model
is
fairly opaque due to its message encapsulation during the
Transfer
(REST)
endpoints traversal which reduces the property of loosely
State
Hierarchical Colored Petri
constructed
for
However, SOAP is well-known to be
transfer
architecture style and Abstract Syntax Notation One XML Encoding Rule (XER) adoption.
environment [3].
message
the
WS extension
coupling among its peers communication [4]. HTTP
allows
a
certain
degree
of
SOAP over
intermediary
for
framework formalization. The validation and verification of
communication agent but decoupling the peers may require
the model is performed by analyzing the generated State Space
extra efforts due to partial message in-transit transparency.
(SS).
Furthermore, the general defacto message format based
It is the aim of this paper as a proof of concept of the larger
research titled "Web Services Extension Model in Colored Petri Net to Enhanced the Message Transfer Performance", an architectural
extension
proposal
to
the
current
remote
laboratory remote instrument connected to the grid for facility sharing aimed at higher communication performance.
on eXtensible Markup Language (XML) is highly verbose and relatively larger in size since the content must be intelligible to human hence inefficient for machine-to machine transport
For message size reduction to
[5].
achieve higher rate of communication,
XML must be
encoded using common standard algorithm understood both I.
at the client-server sides and the communication agents [6].
INTRODUCTION
The need for higher performance data transfer for the
Grid's remote lab toward sharing instrument/sensor within a virtual organization has initiated the motivation of this research area.
The primary objective is web services
extension (WS-E) model in hierarchical Colored Petri Net (HCPN) to the current framework by incorporating Abstract
Syntax Notation One's XML Encoding Rule (ASN.I XER)
Arising from this need, a WS extension (WS-E) to provide alternative solution is proposed.
The extended
framework is studied, formalized, designed, modeled, and analyzed as proof of concept.
The model in HCPN is
verified by analyzing its State Space properties to evaluate
the 'correctness' when simulated using real-world data to validate its performance improvement.
for message size reduction and adopting Representational
III.
State Transfer (REST) architecture style to improve the
RELATED LITERATURE
As part of the current Grid middleware, Globus Tele
loose coupling and scalability properties while supporting transparent intermediary of the service oriented architecture
Control Protocol (GTCP), Common Instrument Middleware
in wide area network. The model simulation replications are
Architecture
performed to obtain the results. State Space of the WS-E
Instrumentation with Distributed Control and Computation
(CIMA)
and
Grid
Enabled
Remote
model is validated and verified for its correctness in terms of
(GridCC), are the choices yet competing efforts in providing
Petri Net token marking's
Grid to remote instrument integration in heterogeneous
discover
the
structural
transitions properties.
liveness and reachability to
interconnection
of
places
and
The aim is to provide the higher
message transfer performance toward heterogeneous data sources
while
promoting
further
interoperability
Coined by [9] REST has gained momentum since then, it
achieve efficient and economical in performing massive computation and conducting virtual remote experimentation between distant sites [I]. The scientific community has the need for higher performance of message transfer from the current of 1�2 cycle per second up to 7�8 folds to perform their experiment using remote lab instrument within the Grid virtual organization adequately [2]. The Grid middleware, such as Globus Toolkit (GT) and gLite, employ common WS framework based on Simple Protocol
Architecture (SOA)
(SOAP)
in
an
architectural
style
(not
a
standard)
to
establish
foundation how the web should be and work. Hypermedia
BACKGROUND
Sharing resources is the essential purpose of the Grid to
Access
attempting to provide standards [8]. is
II.
These instrument middleware (1M) are
based on WS framework for the interoperability while
and
scalability.
Object
environment [7].
Service
Oriented
for interoperability in client-server
utilization,
loose-coupling,
intermediary,
and
uniform
interface are the fundamental desired properties for the web based services [10]. when
fully
According to this architecture style,
supported,
the
web
will
be
interoperable,
scalable, simpler yet fully functional as what the web should be. REST hypermedia engine implies that the resources as services will have unique uniform resources identifications (URis) and the access are performed via links navigation. In other words, RESTful style dictates that web services must expose its resources identified by URI [11]. ASN.I
has
been
around
and
utilized
in
the
telecommunication industry for over than 20 years [12]. It
978-1-4673-1938-6/12/$3\.00 ©2012 IEEE
[741 ]
2012
International Conference on Computer & Information Science (ICCIS)
is a well-proven technology (and algorithm contained) to
simpler term. The random variate value generators provided
produce efficient data transfer at lower layer (hardware
by the CPN tools are:
level). In the last few years ASN.l has gained popularity in the
emerging
internetwork
for
web
services
•
data
uniform(a:real, b:real) : real exponential(r:real) : real
communication. The effort of incorporating ASN.l into web
•
normal(n:real, v:real) : real
services, such as Fast Web Services (FWS), to produce
•
erlang (n:int, r:real) : real
smaller message size while retaining the SOAP has been established [13]. However since SOAP over HTTP does not fully conform and comply with RESTful style, the inherited outstanding issues mentioned previously are still present [14].
In this research, although consideration to utilize the
highest message size reduction such as Packed Encoding Rule (PER) has been investigated, ASN.l
XER is the
middle-ground solution that is still intelligible to human for analysis/debugging
and
easily
processed
format
while
giving the size reduction needed for the Grid's remote instrument message transfer, is the most logical encoding rule to employ [15]. A modeling is the primary approach to realizing the proposed
WS-E.
Henceforth,
bipartite
graph-theory
mathematics based Petri Net is utilized to explain and formulate flow process [16].
Colored Petri Net adds time
property to the 'colored' or attribute of each token's (it is also called marking) [17, 18]. IV.
Henceforth, WS-E model utilizes few of these libraries to provide the set stochastic functions to simulate the real world behavior. The model is constructed from the initial state then all reachable State Space must be calculated.
In order to
properly simulate, the constraint mentioned above must be fulfilled.
The WS-E CPN becomes a reachability graph
with timed values as the colored to the token's marking. All results produced as the outcome of State Space calculations. The Strongly Connected Components (SCC) is the support that each
state
is alive
and reachable while allowing
conditional branching to determine the course of the token traverse according to the marking.
In WS-E HCPN, dead
marking indicated in the State Space report, demonstrate the nondeterministic behavior of the model, i.e. packet loss is allowed and message retransmission from the server is reattempt.
The
State
Space
analysis
of
model-based
verification of finite-state concurrent system involves set of reachable
HIERARCHICAL COLORED PETRI NET
Colored Petri Nets (CPN) is convenient for specifying complex concurrent systems.
markings
(nodes),
occurrences
of
binding
elements (arcs), and occurrence sequences (paths).
The property of CPN has
The
realization in the CPN tool itself is simply the collection of places and transitions connected by arcs.
CPN Tools also
mainly been specified directly in terms of the State Spaces
provides
(SS).
generated State Space graph to find the dead markings (valid
The temporal logics, such as CTL, is useful for
expressing properties of concurrent systems [19].
Primary
utilization of CPN is for analyzing a typically model-driven to allow what-if questions in the form of reachability graph. Typical
CPN
model
types
are
aimed
at
validation,
verification and performance analysis. Petri Net is based on graph theory by possessing the moving token (marking) from places through transitions. events
(transitions)
by
having
a
well-defined
formal
semantics for true concurrency instead of interleaving.
appropriate
functions
for
querying
the
termination states or deadlocks) in order to check the
model's safety and liveness properties. In order to satisfy the statistical minimum requirement to obtain simulation results
[21],
subruns and confidence
intervals must have •
replication of multiple runs since a single run does not
•
mutually
In other words, CPN
provides an explicit representation of states (markings) and
the
provide information about reliability of results independent
runs
to
further
calculate
a
confidence interval, e.g., the flow time is with 95% confidence within the interval 5.5+/-0.5 [5,6].
Simulation is the most commonly used analysis technique and from CPN technical point of view, it is the traverse of
its tokens throughout the reachability graph.
The tokens'
transient behavior, according to the given set of marking (or colors),
enables
indicators.
the
timed
properties
for
performance
It is, however, only an approach method when
the real-world representation is not available. CPN utilizes State Space in its analysis technique; hence, the constraints below must be fulfilled: •
Events or occurrences must finite in sequences, or break-point (or stop-signal) can be employed
•
Reachability enables transition in occurrence sequence
•
Bounded limit must hold upper value for all reachable
to identify deadlock markings The basic of CPN is the stochastic process in which the simulation of nondeterministic system made possible. timed
nondeterministic
probabilities described by
system,
the
In a
durations
and
the given distributions
[20].
Therefore, CPN is to simulate a stochastic process, in
[742 ]
V. WS-E HCPN CONSTRUCTION The WS-E model is designed mainly emphasizing on message
synchrocity,
partial
cacheability,
out-of-order
(nonsequential) message transfer, and data reconstruction at the receiving end.
Concurrency for multiple senders, are
constructed however this research only focuses on dual servers and clients, or receiver, to simplify the initial model. Nondeterministic characteristics due to real-world nature of message transfer is controlled by message guard to ensure the correct sequence order during data reassembly, i.e., message may be delayed, discarded if unsafe, or rearrange when necessary. assigned
to
CPN token color, or time attribute, is
simulate
the
real-life
typical
network
performance. The Hierarchical Colored Petri Net (HCPN) includes additional features such as substitution transitions and fusion places [22].
The underlying idea in HCPN is to use a
number of small CPN in order to develop a complete model. These CPNs are called pages (or subpages) and are related
2012
International Conference on Computer & Information Science (ICCIS)
to each other in a well-defined way.
. "�
:: !::�!::,� ..'_l
��§��:�;g �i§ §�': �:i . ·
���-- ��
� �
..
.QI.".' .,....� • .jj,.", •••••• Ie
::: :';;'::-':':-.'��' ·'"n, ... rQ,.�u,....:;
:�:�;!.:�.: ���,�i:.�: � =
r.::::iI � I
..
\
\
� \
.. "'-\
.',,,J'. ,1,. ." "�. II 11,. • ·O'.: T_·I·T• ..... u ' � : ..:: T lTl
..
'@l
"-.
....... ...
.�;=o::.: ""'···''':>Ct'I«