Improved Strategy for Distributed Processing and Network Application Development
1 like155 views
The document discusses a new software development strategy called 'ee-path' aimed at improving distributed processing and network application development by integrating parallelization, usability, concurrency, and resource distribution. It highlights the shift of performance burden from hardware to software architects due to advancements in multi-core processors, emphasizing a comprehensive approach to both known and unknown user requirements throughout the software lifecycle. The strategy promotes better utilization of resources, recovery from failures, and adaptability to changing requirements, providing a roadmap for software development that includes considerations for future needs.
![International Journal of Computer Applications Technology and Research
Volume 4– Issue 9, 668 - 672, 2015, ISSN: 2319–8656
www.ijcat.com 668
Improved Strategy for Distributed Processing and
Network Application Development
Eke B. O.
Department of Computer Science,
University of Port Harcourt
Port Harcourt, Nigeria
Onuodu F. E.
Department of Computer Science,
University of Port Harcourt
Port Harcourt, Nigeria
Abstract: The complexity of software development abstraction and the new development in multi-core computers have shifted the
burden of distributed software performance from network and chip designers to software architectures and developers. We need to
look at software development strategies that will integrate parallelization of code, concurrency factors, multithreading, distributed
resources allocation and distributed processing. In this paper, a new software development strategy that integrates these factors is
further experimented on parallelism. The strategy is multidimensional aligns distributed conceptualization along a path. This
development strategy mandates application developers to reason along usability, simplicity, resource distribution, parallelization of
code where necessary, processing time and cost factors realignment as well as security and concurrency issues in a balanced path from
the originating point of the network application to its retirement.
Keywords: Parallelization, EE-Path, Distribution, Usability, Concurrency
1. INTRODUCTION
The software strategy referred in this work proffers solution to
distributed software development by using the abstraction of
user requirements and design-time distribution of processes
across multi-core computer powers in multidimensional
visualization, network development, parallelism and
alignment of conceptualization along a path known as the EE-
Path [1]. The technique uses ideas in computational geometry
in trying to resolve a given network, parallelism and
distributed software engineering problem. It is common to see
software specified from the view point of the owners and from
the ideas of similar existing application. It can also be seen
from the point of cost and benefit as well as processing time
and computer resources in a combined or peered manner.
Multi-core computers have shifted the burden of software
performance from chip designers to software architects and
developers. In order to gain the full benefits of this new
hardware, we need to parallelize our code [2]. Parallelization,
therefore, need a design strategy that can guide the software
development process in a distributed system from the
inception to the deployment of the software.
The goal of this paper is to use a development strategy (EE-
Path) that aligns parallelization, usability, distribution, user
requirement abstraction along a balance path during software
development. Since we must overcome software complexity
paradox to achieve the level of simplicity demanded by users
we must think not just along the specified requirements of the
user as classical strategy demand but also on the unspecified
requirements and machine commitment which belong to the
other dimensions in the EE-path. The weakness in the other
strategies is their inability to distribute software design and
development load across processes and processors as well as
the unspecified requirements into the software building plan.
They often ignore or allow programmers to take distribution
and parallelism responsibility. Problems often arise where
programmers depend on the software plan in developing the
system.
2. PARALLELIZATION
Parallelism is a form of computation in which many
calculations are carried out simultaneously, operating on the
principle that large problems can often be divided into smaller
ones, which are then solved concurrently, or ‘in parallel’.
Parallelism is all about decomposing a single task into smaller
ones to enable concurrent execution [2]. Usually, processor
would execute instructions sequentially, which meant that the
vast majority of software was typically written for serial
computation. While we were able to improve the speed of our
processors by increasing the frequency and transistor count, it
was only when computer scientists realized that they had
reached the processor frequency limitation that they started to
explore new methods for improving processor performance
[3]. They explored the use of the germanium in place of
silicon, co-locating many low frequency and power
consuming cores together, adding specialized cores, 3D
transistors, and others. In this era of multi-core processors
exploiting large-scale parallel hardware will be essential for
improving application performance and its capabilities in
terms of executing speed.
Multithreading can be on a single-processor machine, but
parallelism can only occur on a multi-processor machine.
Multiple running threads can be referred to as being concurrent
but not parallel. Concurrency is often used in servers that
operate multiple threads to process requests. However,
parallelism is about decomposing a single task into smaller
ones to enable execution on multiple processors in a
collaborative manner to complete one task. Distributed systems
are a form of parallel computing; however, in distributed
computing, a program is split up into parts that run
simultaneously on multiple computers communicating and
sharing data over a network. By their very nature, distributed
systems must deal with heterogeneous environments, network
links of varying latencies, and unpredictable failures in the
network and the computers.](https://image.slidesharecdn.com/ijcatr04091006-151114075125-lva1-app6892/85/Improved-Strategy-for-Distributed-Processing-and-Network-Application-Development-1-320.jpg)
![International Journal of Computer Applications Technology and Research
Volume 4– Issue 9, 668 - 672, 2015, ISSN: 2319–8656
www.ijcat.com 669
3. THE SOFTWARE DEVELOPMENT
STRATEGY
The EE-Path software development technique aims at
resolving the software development complexity resulting from
improper or lack of provision for the Unknown network,
parallelism and user requirements at the time of the software
specification. An architectural pattern understanding is
complex in terms of the three quality attributes: modifiability,
performance, and availability. Software architects on the other
hand think in terms of architectural patterns [4]. However,
what the architecture needs is a global characterization of
architectural patterns in terms of the factors that affect quality
attribute behaviour so that a software design can be
understood in terms of those quality attributes. Software
engineers must not shay away from complexities of seemingly
intractable parallelism concerns in specifying software
requirement analysis and design.
The quality attributes of architectural patterns are the design
primitives of the software and they are system independent. In
designing software architecture for a product line, the long life
and the flexibility of the software must be of paramount
importance. The full set of requirements of the system is
sparsely known. When the actual products are created there
still remain the Unknowns or better still the unknowable in the
product line. New users may emerge, newer needs may arise
and working environments may change such as operating
system, databases, server changes and machine speed
improvements, multi-core processor changes. These changes
will drive the entire system to reflect the realities of the trend,
creating the need for rapid response to such changes. Our
software development strategy provides a solution to this need
by making architectural provision for the Unknown and also a
room for the Unknown in the entire life cycle of the software.
In areas where parallelism is previously envisaged dummy
checks can be deployed to recover from drawbacks such as
system slowdown, thread race conditions and unforeseen
dependencies.
The Unknown implies all the unspecified requirements of the
software at inception. It also includes all the unforeseen user
need that could give rise to the deployment of parallelism
such as video and heavy image inclusions in software. It
seems that irrespective of the software development method
used, it is the user or software client that specifies what the
software is to do. Irrespective of the way it was specified or
the way the information is collected the target of the software
will depend largely on what the users or software clients
actually want whether they know what they want or not. It is
also true that in most cases the users do not know how to
specify the details of what they want even when they are well
consulted. Some of their specifications are capable of
compromising speed, multi-core processor efficiency and
concurrency. Clients may not be software gurus and may not
specify the software requirement to the extent that all
requirements are covered. Even where all requirements are
covered, external environmental factors such as operating
system changes, network expansion or upgrades and database
upgrades, introduction of new data for processing, new
formulas as well as security loop holes may make the software
vulnerable, and the need to update the software based on the
new requirements may arises. These unforeseen requirements
we generally refer to as the Unknown.
The capturing of the Unknown involves software abstraction
embedded in the conceptual architecture of the system. The
conceptual architecture is one of four different architectures
identified by Hofmeister, Nord and Soni [5]. It describes the
system(s) being designed in terms of the major design
elements and the relationships among them. The EE-Path
strategy determines the balance of the known architectural
drivers, the known environmental factors, the known
parallelism conditions, the known user simplicity factors as
well as all other hidden factors-(Unknown). The software is
then built along this path at least conceptually. A model of the
EE-Path strategy is illustrated in figure 1.
The architectural drivers are the combination of business,
quality and functional requirements that “shape” the
architecture. The known architecture drivers are represented
in the y-axis while the unknown architectural drivers are
represented in its shadow as Architectural drivers 2. Similarly,
other well known parallelism requirement specification are
represented in the z-axis while their unknown is also
represented using its envisaged shadow as Others RS 2. In
analysis, design and construction the EE-Path takes all the
axis into consideration as providing the necessary balance it
requires to remain on its path of move as the software tends to
retirement. The EE-Path will terminate at a point when the
software peters out, but the issue of when this will take place
also throws up another unknown.
Figure 1: The EE-Path Software Strategy Model
Newer upgrades are more likely to surface with additions of
parallelism requirements that were hitherto unknown at the
earlier versions when the software first hit the market. In
integrating parallelism, two types of data parallelism are
considered:
Explicitly Data Parallelism
Implicitly Data Parallelism
In Explicitly Data Parallelism one just plans a loop that
executes in parallel. This can be done by adding OpenMP
Software
Environment
(Known)
EE-Path
Architectural Drivers
(Known)
Architectural Drivers 2
(UnKnown)
Software
Environment 2
(UnKnown)
Others RS 2
(UnKnown)
Parallelism (Known)](https://image.slidesharecdn.com/ijcatr04091006-151114075125-lva1-app6892/85/Improved-Strategy-for-Distributed-Processing-and-Network-Application-Development-2-320.jpg)
![International Journal of Computer Applications Technology and Research
Volume 4– Issue 9, 668 - 672, 2015, ISSN: 2319–8656
www.ijcat.com 670
pragmas around the loop code, or using a parallel algorithm
from Intel® Threading Building Blocks (TBB), from other
source or by developing one.
In Implicitly Data Parallelism one just call some method that
manipulates the data and the infrastructure (i.e. a compiler, a
framework, or the runtime) that is responsible for parallelizing
the work. For instance, the .NET platform provides LINQ
(Language Integrated Query) that allows the use of the
extension methods, and lambda expressions to manipulate the
data like dynamic languages. The following example
demonstrates implicit data manipulation and parallelism:
C# implicit data manipulation using LINQ
string[] students = { “Bartho”,“Yuntho”, “Barry”,”Friday”};
var student = students.Where(p => p.startsWith(“B”));
C# parallel implicit data manipulation using LINQ (Note the
AsParallel method)
string[] students = { “Bartho”, “Yuntho”, “Barry”,”Friday”};
var student = students.AsParallel().Where(p => p.startsWith(“B”));
In language and compiler-based parallelism, the compiler
understands some special keywords to parallelize part of the
code; for example, in OpenMP you can write the following to
parallelize a loop in C++ [6]:
#pragma omp parallelfor
for ( int j = 0; j< max; j++)
{
Num[j] = 1.0;
}
Language and compiler-based parallelism is easy to use
because the majority of the work falls on the compiler. In
library-based parallelism, the programmer should call the
exposed parallel APIs. For example, if one want to parallelize
a for loop in .NET 4 (C#, or VB) a call on For method from
the System.Thread.Parallel class will suffice in C#:
Parallel.For(0,max,j =>
{
Num[j] = 1.0;
});
This method accepts two integers (from, to) and delegates to
the loop body.
4. EE-PATH IN THE APPLICATION
LIFE CYCLE
The EE-Path is a very flexible strategy. It is suited for
complex; highly interactive applications, where very high
integration is required providing good utilization of
underlying hardware within the network and within the multi-
core machine. The strategy promotes reusability of application
components and possibly performance since design
components planned as unknown is reused when the
requirement gets clear. Software requirements have functional
both abstract and concrete, quality and business constraints.
The abstract requirements are used to generate the software
design while the concrete requirements are used to validate
the decisions made as a result of the abstract requirements [7].
The EE-Path remains a guiding path which the requirements
need to follow during the specification. The path is not
introducing any requirement but it provides a structure and a
reference point in the specification of the software
requirements. The use case has the functionality in the system
that gives a user a result of value and captures the functional
requirements [7]. The use case therefore needs to be projected
along the EE-Path to be able to reflect both the known and the
unknown requirements.
Klain [9] believes that the choice of architectural style is
based on the architectural drivers for the design elements that
fit the need at hand. We however believe that architectural
style should not just be based on the need at hand but also on
envisaged need and the unknown future needs. These
unknown needs should be represented using any appropriate
representation in the architecture. Design consideration also
must take into account the specified unknown so that the
unknown can be well specified at least at the abstract
component design level where commitment is yet to be made
to actual software components. The unknown is therefore well
represented in the modular design and aggregated in the
object-oriented class abstraction even if the abstraction is at
worst a dummy. The class abstraction has an inert effect at
making sure some force is exerted to keep the software
development effort on the EE-Path. In the path, the logical,
process, implementation and deployment views are realigned
with the parallelism views even when the software is targeted
at a standalone machine. The standalone can be multi-core
and can equally migrate easily to multi-user when new
requirements surface. This path alignment boosts the
modifiability of the software even when it is already
deployed.
5. DISCUSSION OF EE-PATH BENEFITS
AND PARALLELISM
The EE-Path strategy increases the usability of software since
aggregation is encouraged by patterning one or more actions
on more than one object, even when the object is unknown. It
also makes the system, rather than the user, responsible for
iteration. Furthermore, it is very easy to recover from failure
since the unknown is taking into consideration right from the
architectural stage of the software. Recovery could easily be
based on the unknown functionality of the environment, such
as OS failure and machine failures and even unknown
dependency conditions in parallelized system implementation.
In order to take advantage of the EE-Path in software
development specification for multi-core machines, programs
must be parallelized. Multiple paths of execution have to work
together to complete the tasks the program has to perform,
and that needs to happen concurrently, wherever possible and
in an integrated manner with other requirements. Only then is
it possible to speed up the program . Amdahl’s law expresses
this as [10]:](https://image.slidesharecdn.com/ijcatr04091006-151114075125-lva1-app6892/85/Improved-Strategy-for-Distributed-Processing-and-Network-Application-Development-3-320.jpg)
![International Journal of Computer Applications Technology and Research
Volume 4– Issue 9, 668 - 672, 2015, ISSN: 2319–8656
www.ijcat.com 671
Figure 2 : Illustrating Speedup and Number of processors
where S is the speed-up of the program (as a factor of its
original sequential runtime), and P is the fraction that is
parallelizable.
Determining when and where in the software to inject
parallelism is a challenge and if wrongly decided could have
retrogressive consequences hence most of decision could be
provided as unknown at certain stage of the system. The good
thing in the EE-Path is that strong provision is made for its
implementation at worst as a dummy implementation. This
will help developers to plan ahead even when it is not feasible
to implement it at the earlier releases of the software. One
probably do not need to parallelize if the application is really
simple and the code is running fast enough already. But we
know that a simple application today may turn to a complex
application with time and a fast code could slow down when
new users use it in a network or when newer features are
added hence the need to plan for the unknown via the EE-
Path. Network applications use shared data, hence
dependency issues can be planned using the EE-Path to take
other factors into considerations to avoid pitfalls of delays as a
result of dependency and thread locks. Decisions when made
at the planning stage guides developers in the choice of
parallel frameworks and APIs to use during the application
implementation. These will help in leveraging the power of all
the extra cores on developers and users machines. The EE-
Path strategy encourages developers to leverage their
knowledge and also to develop systems in relatively
unfamiliar parallelized contexts as offered by distributed
application environments. The unknown is not fixed but it
remains the unknown as long as it has not been unraveled and
since no human can have full and final insight of any matter at
any given time, progressive development is encouraged by
our strategy.
6. OUR CONTRIBUTIONS.
In this paper, we incorporate a new software strategy which is
able to implement a multidimensional requirement
visualization of three or more lines of simultaneous
requirement alignment. It inculcates the unspecified
requirement that we see as forming the core of modern system
design and allow parallelism requirement analysis and design
to varying level of implementation. When the requirement is
not needed at the moment we postulate it can be allowed to be
implemented as an abstract class in the system without any
derivation or with dummy derivation. Some of the
requirement issues to be considered include security,
concurrency, interoperability, reusability and the Unknown.
The Unknown class can be specified with all possible
abstraction that can be modified in the future when the need
for the Unknown requirement arises. This design technique
takes care of the Unknown making the system to be
extendible without the need to redesign the system. This
design technique takes care of the light-speed changes in
requirements resulting in the development of newer versions
of software within very short period of time ranging from few
days to few months. It breaks parallelism conditions in the
software requirement to determine where it can be
implemented to maximize speed. It also articulates pitfalls to
avoid deployment of parallelism to those areas where
parallelism could lead to processing slow-down or incorrect
generation of result. There are many parallelism frameworks,
and debugging tools aimed at simplifying the task of parallel
programming, such as:
Intel Parallel Studio, Microsoft CCR and DSS, MS PPL -
Microsoft Parallel Pattern Library (was released in 2009 Q4),
MS .NET 4 - Microsoft .NET Framework 4 (will released in
2009 Q4), Java 7 (will release in 2009), PRL - Parallel
Runtime Library (Beta 1 released in June 2009) [2]. Software
engineers need to integrate this entire requirement in system
development early in the system life-cycle while making
provision for the unknown.
7. CONCLUSION
The EE-Path software development strategy provides a means
of guiding developers and software architects in qualitative
measures of marginal building blocks in choosing and
developing architectural styles and in conceptualization of the
system at hand from the inception to the conclusion. Based on
the evaluation of software complexity and other models it can
be seen that if the EE-path is followed, a better preparation for
the unknown is made. Furthermore, it can be seen that for the
parallelization of network application the EE-Path model
offers variables for consideration and integration of other
factors and requirements in the development of software in a
hitherto different network platforms. These provide improved
standardization of development even at the architectural level
of software development. It can therefore be concluded that
developing network software using the EE-Path concept
results in building a software today with provision made for
change which itself appears to be a constant in the world of
software engineering.
8. REFERENCES
[1] Eke B. O. and Nwachukwu E. O. (2011), Software
Engineering Process: Yaam Deployment in E-
Bookshop Use Case Scenario, Journal of Theoretical
and Applied Information Technology, Vol 30 No. 2
August, 2011, JATIT and LLS, www.jatit.org E-
ISSN:1817-3195, ISSN:1992-8645, Islamabad,Pakistan,
http://www.jatit.org/volumes/Vol30No2/2Vol30No2.pdf](https://image.slidesharecdn.com/ijcatr04091006-151114075125-lva1-app6892/85/Improved-Strategy-for-Distributed-Processing-and-Network-Application-Development-4-320.jpg)
![International Journal of Computer Applications Technology and Research
Volume 4– Issue 9, 668 - 672, 2015, ISSN: 2319–8656
www.ijcat.com 672
[2] Fadeel, H. (2009) Getting Started with Parallel
Programming, http://library.dzone.com/category/
Accessed, June 2015.
[3] James Reinders (2009) Getting Started With Intel
Threading Building Blocks, http://library.dzone.com
Accessed July 2009
[4] Katzman, R.; Barbacci, M.; Carriere S. J.; and Woods, S.
J. (2014) Experience with Performing Architectural Styles,
Software Architecture Tradeoff Analysis. 54-63. Proceedings
of ICSE99. Los Angeles, CA, May 1999.
[5] Hofmeister, C.; Nord, R.; and Soni, P. (2000) Applied
Software Architecture. Reading MA: Addison Wesley.
[6] James P. C.; Jack W. D.; C++ Program Design An
Introduction to Programming and Object-Oriented Design
WCB McGraw-Hill USA
[7] Bass L.; Klein M.; Bachman F. (2000) Quality Attribute
Design Primitives (CMU/SEI-2000IN-2000-017). Pittsburgh,
PA: Software Engineering Institute, Carnegie Mellon
University.
[8] Jacobson, I.(1992); Object-Oriented Software
Engineering, Addison-Wesley, USA
[9] Klein M.; Kazman R.; Barbacci M.; Carriere S.;and
Lipson, H.(1999) Attribute-Based Architectural Styles,
Software Architecture. 225-243. Proceedings of the First
Working IFIP Conference on Software Architecture
(WICSA1), San Antonio, TX, February 1999.
[10] Wikipedia, Parallel computing,
www.en.wikipedia.org/wiki/parallel_computing accessed July
2014
7. ACKNOWLEDGMENTS
Our thanks to the Oyol Computer Consult Inc and Fonglo
Research Center for their contribution in typesetting and
towards development of the work.
8. ABOUT THE AUTHORS
Eke Bartholomew PhD, MCPN,
MACM, FIPMD is a Software
Engineering / Computer
Science Lecturer at the
University of Port Harcourt and
Mobile Application Developer
in Oyol Computer Consult Inc.
His research interest is in SE
Methodologies and Mobile IT
deployment.
Dr. Onuodu, Friday E. is a
Lecturer at the University of
Port Harcourt. His research
interest is in Data Mining and
Data Extraction using both
Mobile Devices and Desktops.
He also has interest in IT
deployment analysis. He has
many publication in learned
journals.](https://image.slidesharecdn.com/ijcatr04091006-151114075125-lva1-app6892/85/Improved-Strategy-for-Distributed-Processing-and-Network-Application-Development-5-320.jpg)
Improved Strategy for Distributed Processing and Network Application Development
- 1. International Journal of Computer Applications Technology and Research Volume 4– Issue 9, 668 - 672, 2015, ISSN: 2319–8656 www.ijcat.com 668 Improved Strategy for Distributed Processing and Network Application Development Eke B. O. Department of Computer Science, University of Port Harcourt Port Harcourt, Nigeria Onuodu F. E. Department of Computer Science, University of Port Harcourt Port Harcourt, Nigeria Abstract: The complexity of software development abstraction and the new development in multi-core computers have shifted the burden of distributed software performance from network and chip designers to software architectures and developers. We need to look at software development strategies that will integrate parallelization of code, concurrency factors, multithreading, distributed resources allocation and distributed processing. In this paper, a new software development strategy that integrates these factors is further experimented on parallelism. The strategy is multidimensional aligns distributed conceptualization along a path. This development strategy mandates application developers to reason along usability, simplicity, resource distribution, parallelization of code where necessary, processing time and cost factors realignment as well as security and concurrency issues in a balanced path from the originating point of the network application to its retirement. Keywords: Parallelization, EE-Path, Distribution, Usability, Concurrency 1. INTRODUCTION The software strategy referred in this work proffers solution to distributed software development by using the abstraction of user requirements and design-time distribution of processes across multi-core computer powers in multidimensional visualization, network development, parallelism and alignment of conceptualization along a path known as the EE- Path [1]. The technique uses ideas in computational geometry in trying to resolve a given network, parallelism and distributed software engineering problem. It is common to see software specified from the view point of the owners and from the ideas of similar existing application. It can also be seen from the point of cost and benefit as well as processing time and computer resources in a combined or peered manner. Multi-core computers have shifted the burden of software performance from chip designers to software architects and developers. In order to gain the full benefits of this new hardware, we need to parallelize our code [2]. Parallelization, therefore, need a design strategy that can guide the software development process in a distributed system from the inception to the deployment of the software. The goal of this paper is to use a development strategy (EE- Path) that aligns parallelization, usability, distribution, user requirement abstraction along a balance path during software development. Since we must overcome software complexity paradox to achieve the level of simplicity demanded by users we must think not just along the specified requirements of the user as classical strategy demand but also on the unspecified requirements and machine commitment which belong to the other dimensions in the EE-path. The weakness in the other strategies is their inability to distribute software design and development load across processes and processors as well as the unspecified requirements into the software building plan. They often ignore or allow programmers to take distribution and parallelism responsibility. Problems often arise where programmers depend on the software plan in developing the system. 2. PARALLELIZATION Parallelism is a form of computation in which many calculations are carried out simultaneously, operating on the principle that large problems can often be divided into smaller ones, which are then solved concurrently, or ‘in parallel’. Parallelism is all about decomposing a single task into smaller ones to enable concurrent execution [2]. Usually, processor would execute instructions sequentially, which meant that the vast majority of software was typically written for serial computation. While we were able to improve the speed of our processors by increasing the frequency and transistor count, it was only when computer scientists realized that they had reached the processor frequency limitation that they started to explore new methods for improving processor performance [3]. They explored the use of the germanium in place of silicon, co-locating many low frequency and power consuming cores together, adding specialized cores, 3D transistors, and others. In this era of multi-core processors exploiting large-scale parallel hardware will be essential for improving application performance and its capabilities in terms of executing speed. Multithreading can be on a single-processor machine, but parallelism can only occur on a multi-processor machine. Multiple running threads can be referred to as being concurrent but not parallel. Concurrency is often used in servers that operate multiple threads to process requests. However, parallelism is about decomposing a single task into smaller ones to enable execution on multiple processors in a collaborative manner to complete one task. Distributed systems are a form of parallel computing; however, in distributed computing, a program is split up into parts that run simultaneously on multiple computers communicating and sharing data over a network. By their very nature, distributed systems must deal with heterogeneous environments, network links of varying latencies, and unpredictable failures in the network and the computers.
- 2. International Journal of Computer Applications Technology and Research Volume 4– Issue 9, 668 - 672, 2015, ISSN: 2319–8656 www.ijcat.com 669 3. THE SOFTWARE DEVELOPMENT STRATEGY The EE-Path software development technique aims at resolving the software development complexity resulting from improper or lack of provision for the Unknown network, parallelism and user requirements at the time of the software specification. An architectural pattern understanding is complex in terms of the three quality attributes: modifiability, performance, and availability. Software architects on the other hand think in terms of architectural patterns [4]. However, what the architecture needs is a global characterization of architectural patterns in terms of the factors that affect quality attribute behaviour so that a software design can be understood in terms of those quality attributes. Software engineers must not shay away from complexities of seemingly intractable parallelism concerns in specifying software requirement analysis and design. The quality attributes of architectural patterns are the design primitives of the software and they are system independent. In designing software architecture for a product line, the long life and the flexibility of the software must be of paramount importance. The full set of requirements of the system is sparsely known. When the actual products are created there still remain the Unknowns or better still the unknowable in the product line. New users may emerge, newer needs may arise and working environments may change such as operating system, databases, server changes and machine speed improvements, multi-core processor changes. These changes will drive the entire system to reflect the realities of the trend, creating the need for rapid response to such changes. Our software development strategy provides a solution to this need by making architectural provision for the Unknown and also a room for the Unknown in the entire life cycle of the software. In areas where parallelism is previously envisaged dummy checks can be deployed to recover from drawbacks such as system slowdown, thread race conditions and unforeseen dependencies. The Unknown implies all the unspecified requirements of the software at inception. It also includes all the unforeseen user need that could give rise to the deployment of parallelism such as video and heavy image inclusions in software. It seems that irrespective of the software development method used, it is the user or software client that specifies what the software is to do. Irrespective of the way it was specified or the way the information is collected the target of the software will depend largely on what the users or software clients actually want whether they know what they want or not. It is also true that in most cases the users do not know how to specify the details of what they want even when they are well consulted. Some of their specifications are capable of compromising speed, multi-core processor efficiency and concurrency. Clients may not be software gurus and may not specify the software requirement to the extent that all requirements are covered. Even where all requirements are covered, external environmental factors such as operating system changes, network expansion or upgrades and database upgrades, introduction of new data for processing, new formulas as well as security loop holes may make the software vulnerable, and the need to update the software based on the new requirements may arises. These unforeseen requirements we generally refer to as the Unknown. The capturing of the Unknown involves software abstraction embedded in the conceptual architecture of the system. The conceptual architecture is one of four different architectures identified by Hofmeister, Nord and Soni [5]. It describes the system(s) being designed in terms of the major design elements and the relationships among them. The EE-Path strategy determines the balance of the known architectural drivers, the known environmental factors, the known parallelism conditions, the known user simplicity factors as well as all other hidden factors-(Unknown). The software is then built along this path at least conceptually. A model of the EE-Path strategy is illustrated in figure 1. The architectural drivers are the combination of business, quality and functional requirements that “shape” the architecture. The known architecture drivers are represented in the y-axis while the unknown architectural drivers are represented in its shadow as Architectural drivers 2. Similarly, other well known parallelism requirement specification are represented in the z-axis while their unknown is also represented using its envisaged shadow as Others RS 2. In analysis, design and construction the EE-Path takes all the axis into consideration as providing the necessary balance it requires to remain on its path of move as the software tends to retirement. The EE-Path will terminate at a point when the software peters out, but the issue of when this will take place also throws up another unknown. Figure 1: The EE-Path Software Strategy Model Newer upgrades are more likely to surface with additions of parallelism requirements that were hitherto unknown at the earlier versions when the software first hit the market. In integrating parallelism, two types of data parallelism are considered: Explicitly Data Parallelism Implicitly Data Parallelism In Explicitly Data Parallelism one just plans a loop that executes in parallel. This can be done by adding OpenMP Software Environment (Known) EE-Path Architectural Drivers (Known) Architectural Drivers 2 (UnKnown) Software Environment 2 (UnKnown) Others RS 2 (UnKnown) Parallelism (Known)
- 3. International Journal of Computer Applications Technology and Research Volume 4– Issue 9, 668 - 672, 2015, ISSN: 2319–8656 www.ijcat.com 670 pragmas around the loop code, or using a parallel algorithm from Intel® Threading Building Blocks (TBB), from other source or by developing one. In Implicitly Data Parallelism one just call some method that manipulates the data and the infrastructure (i.e. a compiler, a framework, or the runtime) that is responsible for parallelizing the work. For instance, the .NET platform provides LINQ (Language Integrated Query) that allows the use of the extension methods, and lambda expressions to manipulate the data like dynamic languages. The following example demonstrates implicit data manipulation and parallelism: C# implicit data manipulation using LINQ string[] students = { “Bartho”,“Yuntho”, “Barry”,”Friday”}; var student = students.Where(p => p.startsWith(“B”)); C# parallel implicit data manipulation using LINQ (Note the AsParallel method) string[] students = { “Bartho”, “Yuntho”, “Barry”,”Friday”}; var student = students.AsParallel().Where(p => p.startsWith(“B”)); In language and compiler-based parallelism, the compiler understands some special keywords to parallelize part of the code; for example, in OpenMP you can write the following to parallelize a loop in C++ [6]: #pragma omp parallelfor for ( int j = 0; j< max; j++) { Num[j] = 1.0; } Language and compiler-based parallelism is easy to use because the majority of the work falls on the compiler. In library-based parallelism, the programmer should call the exposed parallel APIs. For example, if one want to parallelize a for loop in .NET 4 (C#, or VB) a call on For method from the System.Thread.Parallel class will suffice in C#: Parallel.For(0,max,j => { Num[j] = 1.0; }); This method accepts two integers (from, to) and delegates to the loop body. 4. EE-PATH IN THE APPLICATION LIFE CYCLE The EE-Path is a very flexible strategy. It is suited for complex; highly interactive applications, where very high integration is required providing good utilization of underlying hardware within the network and within the multi- core machine. The strategy promotes reusability of application components and possibly performance since design components planned as unknown is reused when the requirement gets clear. Software requirements have functional both abstract and concrete, quality and business constraints. The abstract requirements are used to generate the software design while the concrete requirements are used to validate the decisions made as a result of the abstract requirements [7]. The EE-Path remains a guiding path which the requirements need to follow during the specification. The path is not introducing any requirement but it provides a structure and a reference point in the specification of the software requirements. The use case has the functionality in the system that gives a user a result of value and captures the functional requirements [7]. The use case therefore needs to be projected along the EE-Path to be able to reflect both the known and the unknown requirements. Klain [9] believes that the choice of architectural style is based on the architectural drivers for the design elements that fit the need at hand. We however believe that architectural style should not just be based on the need at hand but also on envisaged need and the unknown future needs. These unknown needs should be represented using any appropriate representation in the architecture. Design consideration also must take into account the specified unknown so that the unknown can be well specified at least at the abstract component design level where commitment is yet to be made to actual software components. The unknown is therefore well represented in the modular design and aggregated in the object-oriented class abstraction even if the abstraction is at worst a dummy. The class abstraction has an inert effect at making sure some force is exerted to keep the software development effort on the EE-Path. In the path, the logical, process, implementation and deployment views are realigned with the parallelism views even when the software is targeted at a standalone machine. The standalone can be multi-core and can equally migrate easily to multi-user when new requirements surface. This path alignment boosts the modifiability of the software even when it is already deployed. 5. DISCUSSION OF EE-PATH BENEFITS AND PARALLELISM The EE-Path strategy increases the usability of software since aggregation is encouraged by patterning one or more actions on more than one object, even when the object is unknown. It also makes the system, rather than the user, responsible for iteration. Furthermore, it is very easy to recover from failure since the unknown is taking into consideration right from the architectural stage of the software. Recovery could easily be based on the unknown functionality of the environment, such as OS failure and machine failures and even unknown dependency conditions in parallelized system implementation. In order to take advantage of the EE-Path in software development specification for multi-core machines, programs must be parallelized. Multiple paths of execution have to work together to complete the tasks the program has to perform, and that needs to happen concurrently, wherever possible and in an integrated manner with other requirements. Only then is it possible to speed up the program . Amdahl’s law expresses this as [10]:
- 4. International Journal of Computer Applications Technology and Research Volume 4– Issue 9, 668 - 672, 2015, ISSN: 2319–8656 www.ijcat.com 671 Figure 2 : Illustrating Speedup and Number of processors where S is the speed-up of the program (as a factor of its original sequential runtime), and P is the fraction that is parallelizable. Determining when and where in the software to inject parallelism is a challenge and if wrongly decided could have retrogressive consequences hence most of decision could be provided as unknown at certain stage of the system. The good thing in the EE-Path is that strong provision is made for its implementation at worst as a dummy implementation. This will help developers to plan ahead even when it is not feasible to implement it at the earlier releases of the software. One probably do not need to parallelize if the application is really simple and the code is running fast enough already. But we know that a simple application today may turn to a complex application with time and a fast code could slow down when new users use it in a network or when newer features are added hence the need to plan for the unknown via the EE- Path. Network applications use shared data, hence dependency issues can be planned using the EE-Path to take other factors into considerations to avoid pitfalls of delays as a result of dependency and thread locks. Decisions when made at the planning stage guides developers in the choice of parallel frameworks and APIs to use during the application implementation. These will help in leveraging the power of all the extra cores on developers and users machines. The EE- Path strategy encourages developers to leverage their knowledge and also to develop systems in relatively unfamiliar parallelized contexts as offered by distributed application environments. The unknown is not fixed but it remains the unknown as long as it has not been unraveled and since no human can have full and final insight of any matter at any given time, progressive development is encouraged by our strategy. 6. OUR CONTRIBUTIONS. In this paper, we incorporate a new software strategy which is able to implement a multidimensional requirement visualization of three or more lines of simultaneous requirement alignment. It inculcates the unspecified requirement that we see as forming the core of modern system design and allow parallelism requirement analysis and design to varying level of implementation. When the requirement is not needed at the moment we postulate it can be allowed to be implemented as an abstract class in the system without any derivation or with dummy derivation. Some of the requirement issues to be considered include security, concurrency, interoperability, reusability and the Unknown. The Unknown class can be specified with all possible abstraction that can be modified in the future when the need for the Unknown requirement arises. This design technique takes care of the Unknown making the system to be extendible without the need to redesign the system. This design technique takes care of the light-speed changes in requirements resulting in the development of newer versions of software within very short period of time ranging from few days to few months. It breaks parallelism conditions in the software requirement to determine where it can be implemented to maximize speed. It also articulates pitfalls to avoid deployment of parallelism to those areas where parallelism could lead to processing slow-down or incorrect generation of result. There are many parallelism frameworks, and debugging tools aimed at simplifying the task of parallel programming, such as: Intel Parallel Studio, Microsoft CCR and DSS, MS PPL - Microsoft Parallel Pattern Library (was released in 2009 Q4), MS .NET 4 - Microsoft .NET Framework 4 (will released in 2009 Q4), Java 7 (will release in 2009), PRL - Parallel Runtime Library (Beta 1 released in June 2009) [2]. Software engineers need to integrate this entire requirement in system development early in the system life-cycle while making provision for the unknown. 7. CONCLUSION The EE-Path software development strategy provides a means of guiding developers and software architects in qualitative measures of marginal building blocks in choosing and developing architectural styles and in conceptualization of the system at hand from the inception to the conclusion. Based on the evaluation of software complexity and other models it can be seen that if the EE-path is followed, a better preparation for the unknown is made. Furthermore, it can be seen that for the parallelization of network application the EE-Path model offers variables for consideration and integration of other factors and requirements in the development of software in a hitherto different network platforms. These provide improved standardization of development even at the architectural level of software development. It can therefore be concluded that developing network software using the EE-Path concept results in building a software today with provision made for change which itself appears to be a constant in the world of software engineering. 8. REFERENCES [1] Eke B. O. and Nwachukwu E. O. (2011), Software Engineering Process: Yaam Deployment in E- Bookshop Use Case Scenario, Journal of Theoretical and Applied Information Technology, Vol 30 No. 2 August, 2011, JATIT and LLS, www.jatit.org E- ISSN:1817-3195, ISSN:1992-8645, Islamabad,Pakistan, http://www.jatit.org/volumes/Vol30No2/2Vol30No2.pdf
- 5. International Journal of Computer Applications Technology and Research Volume 4– Issue 9, 668 - 672, 2015, ISSN: 2319–8656 www.ijcat.com 672 [2] Fadeel, H. (2009) Getting Started with Parallel Programming, http://library.dzone.com/category/ Accessed, June 2015. [3] James Reinders (2009) Getting Started With Intel Threading Building Blocks, http://library.dzone.com Accessed July 2009 [4] Katzman, R.; Barbacci, M.; Carriere S. J.; and Woods, S. J. (2014) Experience with Performing Architectural Styles, Software Architecture Tradeoff Analysis. 54-63. Proceedings of ICSE99. Los Angeles, CA, May 1999. [5] Hofmeister, C.; Nord, R.; and Soni, P. (2000) Applied Software Architecture. Reading MA: Addison Wesley. [6] James P. C.; Jack W. D.; C++ Program Design An Introduction to Programming and Object-Oriented Design WCB McGraw-Hill USA [7] Bass L.; Klein M.; Bachman F. (2000) Quality Attribute Design Primitives (CMU/SEI-2000IN-2000-017). Pittsburgh, PA: Software Engineering Institute, Carnegie Mellon University. [8] Jacobson, I.(1992); Object-Oriented Software Engineering, Addison-Wesley, USA [9] Klein M.; Kazman R.; Barbacci M.; Carriere S.;and Lipson, H.(1999) Attribute-Based Architectural Styles, Software Architecture. 225-243. Proceedings of the First Working IFIP Conference on Software Architecture (WICSA1), San Antonio, TX, February 1999. [10] Wikipedia, Parallel computing, www.en.wikipedia.org/wiki/parallel_computing accessed July 2014 7. ACKNOWLEDGMENTS Our thanks to the Oyol Computer Consult Inc and Fonglo Research Center for their contribution in typesetting and towards development of the work. 8. ABOUT THE AUTHORS Eke Bartholomew PhD, MCPN, MACM, FIPMD is a Software Engineering / Computer Science Lecturer at the University of Port Harcourt and Mobile Application Developer in Oyol Computer Consult Inc. His research interest is in SE Methodologies and Mobile IT deployment. Dr. Onuodu, Friday E. is a Lecturer at the University of Port Harcourt. His research interest is in Data Mining and Data Extraction using both Mobile Devices and Desktops. He also has interest in IT deployment analysis. He has many publication in learned journals.