Main Features of Networks and Distributed Systems - Coursework Example

Networks and Distributed Systems Table of Contents Table of Contents 2 Introduction 3 Challenges for a Distributed System 4 Fallacies of Distributed Computing 4 Reasons for Failure of a Commercially Distributed Computer System and Relation with Peter Deutshs Fallacies 5 Conclusion 11 References 12 Introduction A distributed system can be described as a compilation of autonomous computers which appears to the users of a system as a single as well as coherent system. A distributed system has been a major reckoning issue in the past two decades. In this period, along with continuous progression in the communication technology and also with the accessibility of powerful yet cheap microprocessors, in PCs, PDAs, Workstations, and embedded systems, the distributed system has been used. Example of distributed system are World Wide Web based information and also resource sharing, network constituted workstations, clusters, network of branch based computers like information system for handling automatic method of processing of any order, distributed manufacturing system like automatic assembly line, Play Station 3 related new cell processors, network based embedded system. A distributed system has various advantages such as speed, economics, reliability, inherent distribution, and incremental growth. The disadvantages are network, security, software, and there are more components which have the possibility to fail. A distributed system tries to achieve various goals such as openness, transparency, performance, reliability, and scalability. There are various types of distributed system such as distributed information systems, distributed computing systems, distributed pervasive or embedded systems (Lee, 2007). Challenges for a Distributed System A distributed system has been increasingly used in commercial use. Distributed system has made it much easier for any organisation to operate different tasks at much easier way and also in a much quicker amount of time. Distributed system though faces certain challenges in its commercial use which can cause a failure in the entire system. These challenges are fault tolerance, secure communication by using public based networks, replication, naming and caching, and also shared state and coordination (Lee, 2007). Fallacies of Distributed Computing In the year 1994, a ‘Sun’ fellow named Peter Deutsch outlined seven different assumptions which the designers and the architects of a distributed system is expected to make. These assumptions turn out to be wrong ultimately resulting in various kinds of pains and troubles for the architects who had considered the assumptions and solutions. In the year 1997, James Gosling contributed to the existing seven fallacies and added another fallacy. These assumptions are named together as ‘The 8 fallacies of Distributed Computing’. These fallacies are: 1. “The network is reliable”. 2. “Bandwidth is infinite”. 3. “Latency is zero”. 4. “Topology doesnt change”. 5. “The network is secure”. 6. “There is one administrator”. 7. “The network is homogeneous”. 8. “Transport cost is zero“. (Rotem-Gal-Oz, n.d.). Reasons for Failure of a Commercially Distributed Computer System and Relation with Peter Deutshs Fallacies A distribution system has added certain benefits to the process of information and resource sharing. These benefits such as distribution, performance, reliability i.e. fault tolerance, sharing of resources and data, incremental growth, and human to human communication have made distributed system a popular use in the commercial purpose. There are also certain difficulties that a distributed system faces like networking problem which can cause loss of data, overloading; loss of messages is created by fault in the networking infrastructure. The other problem lies in the difficulty to create distributed software. A major issue related to a distributed system is security problems i.e. problem related to data security. The issue that can arise in a distributed system because of design related reasons are communication, transparency, heterogeneity, performance and scalability, openness, security, and fault tolerance and reliability (Eles, n.d.). In the process of developing a distributed system for the first time certain fallacies are developed by the first time developing team. These 8 fallacies are very important to analyse a distributed system and also to recognise the failure of a system. A commercial distributed system can be used in various processes such as data processing. In order to make such a commercial distributed system work in a data processing scenario, certain areas needs to work properly like the structure of the distributed system, a large range of scalability, utilisation of extraordinarily parallel computation, security, and high level of system availability (Wilkes, 1986). CORBA (Common Object Request Broker Architecture) is basically a standard architecture used in distributed object oriented systems. This system permits a heterogeneous and also a distributed compilation of substance to interoperate. CORBA provides certain distributed applications like data used are distributed, computation process is also distributed, and users also can work through a distributed network. In the distributed system, the application developers have to take into consideration issues such as communication process between different machines is slower than in a machine located in the same location. In the scenario of distributed system objects generally fail separately. In the distributed system a network can be partitioned and both the objects can be executed independently if the other object fails to operate. In a distributed system concurrent access of data can occur. Distributed system though is certainly not secure and security related issue is of paramount importance in the development of this system (Oracle, 1999). CORBA as a distributed system was constructed with written codes and was prepared as software component. CORBA had to suffer certain failures because of the creation process of CORBA and also because of the implementation reasons. The failures in CORBA led to a decline in the use of CORBA for commercial and other organisational purposes. The main reason for failure CORBA was incompatibilities related to implementation, transparency of the location, lacking in the designing and process related segment, implementation related difficulties, and security related issues like Firewalls. CORBA was traditionally prepared by focusing on interoperability and portability. A specification of CORBA was first completed in the year 1991. In this specification, these two features of standardisation were not provided. In present scenario, the two standardisation features have been included in CORBA but at a limited level (Chappell, 1998). CORBA was initially developed as an early adopted technology. It gradually developed to become a middleware and a niche based technology which is used very scarcely. In the beginning on 1990s communicating between different machines was very difficult. There was need for developing a middleware for making the communication process easier. CORBA was developed provide facilities for making communication faster and easier. CORBA 1.0 was first prepared but did not feature interoperability and featured only C based mapping. In the next version CORBA 2.0 it featured Java and C++ language based mapping. By using this process heterogeneous distributed based application preparation was much easier. During the growth phase of CORBA there was advent of Web and also Java. CORBA included Java language based mapping but it did not cooperate along with the Web. Companies searched for separate solution instead of looking for CORBA to come up with one. CORBA component model was developed to replace the existing model based on common business object facility. The implementation process of component based model was not performed properly thus it transformed to be termed as a ‘failure’. After this failure with CORBA CGI, HTML, and HTTP were used to build distributed systems as there was a real need for a middleware in the industry. The cost of CORBA implementation commercially had cost oriented issues. As the cost of such implementation was thousands of dollars for per development related seat. There was also a compatibility related issue with CORBA as it worked along only with the Windows operating system. Microsoft the developer of Windows operating system promoted their own middleware component Distributed Component Object Model (DCOM) instead of CORBA. CORBA’s failure was also contributed by XML. CORBA is mostly used to connect together different components of a companies networks. A firewall is employed in this scenario to increase the security aspect. CORBA is also used in embedded as well as real-time system development. CORBA use in present scenario is subjected to only in case of niche technologies. Technical excellence is needed in a technology to sustain itself in the long run and also to tackle global competition. CORBA lacked the needed prerequisite of technical excellence which contributed in the failure of CORBA. Complexity was a technical problem with CORBA. CORBA did not provide sufficient features for two main components i.e. security and versioning. In case of commercial e-commerce based infrastructure lack of versioning and security absolutely necessary components. Therefore CORBA got rejected. Flaws in the design of CORBA, ignoring of .NET features were also a part of CORBA’s failure (Henning, 2006). Reliability is also an issue related described in Deutsch’s fallacies. Distributed systems are planned to become the backbone of the developing and emerging generations’ communication system. E-commerce, real-time data related feeds, PCs, satellite based surveillance system, distributed medical related imaging, flight reservation systems are all components of distributed system. Distributed system is used generally to improve quality related to a service offering like availability, performance and reliability. In a distribution system, a partial or initial failure is an inherent occurring problem. This makes attaining reliability much more difficult. In the Deutsch’s fallacies, security is also an aspect. Security features in CORBA include authorisation, identification, authentication, access control, non-repudiation, security auditing, and also administration. The administration aspect is also considered in the fallacy. Identification and authentication aspect in CORBA tries to identify and verify user identity. The administration of security in CORBA is done by using a domain (Moreno, 2002). Transparency is the other issue considered in the fallacy. In case of CORBA, a location based transparency is taken as a notion. This scenario creates a problem in case of requirement of a local machine access. This aspect is discussed in the fallacy ‘topology doesn’t change’. CORBA mainly employs TCP/IP based connections to send out any data. A firewall can create a problem in the data transmission process by creating restriction. By using CORBA user has to face up with such difficulties. Therefore, CORBA’s use became much lesser. Web based services became a preferred choice in the place of CORBA. Due to the interconnected feature of modern computers, authentications along with security are the two extremely important components of distributed system. The aspect of network being termed as homogeneous fallacy, it generally does not get considered in present day scenario as a home based network comprises different components made by separate vendors. Homogeneous network has become an exception in modern day scenario. The transport cost related fallacy can be described as fallacy which relates to latency factor being zero. There are additional costs considered in case of resources and time (Rotem-Gal-Oz, n.d.). Conclusion CORBA as a distributed system has gone through transformation from being considered as a crucial part of development in the field of e-commerce based application to merely transpiring to an only being considered as a niche technology. The fallacies developed by Peter Deutsch are a part of developing distributed system. In the initial stage of a distributed system, there remains a major chance of failure. Therefore, if these fallacies are considered at the outset in the process of developing a distributed system then there would exist reduced chance of meeting with failures. These fallacies generally turn out to be false in the long run. Thus, it generally helps in the initiation process to assume these fallacies as in the end these are going to be proved wrong. Thus, fallacies can help a distributed system to work properly. References Chappell, D., 1998. The Trouble With CORBA. Articles. [Online] Available at: http://www.davidchappell.com/articles/article_Trouble_CORBA.html [Accessed July 29, 2010]. Eles, P., No Date. Distributed Systems. Linköping University. [Online] Available at: http://www.ida.liu.se/~TDDB37/lecture-notes/lect1.frm.pdf [Accessed July 29, 2010]. Henning, M., 2006. The Rise and Fall of CORBA. Acmqueue. [Online] Available at: http://queue.acm.org/detail.cfm?id=1142044 [Accessed July 29, 2010]. Lee, I., 2007. Introduction to Distributed Systems. University of Pennsylvania. [Online] Available at: http://www.cis.upenn.edu/~lee/07cis505/Lec/lec-ch1-DistSys-v4.pdf [Accessed July 29, 2010]. Moreno, A. L., 2002. Distributed Systems Security: Java, CORBA, and COM+. SANS Institute. [Online] Available at: http://www.sans.org/reading_room/whitepapers/application/distributed-systems-security-java-corba-com-plus_28 [Accessed July 29, 2010]. Oracle, 1999. Introduction to CORBA. Sun Developer Network. [Online] Available at: http://java.sun.com/developer/onlineTraining/corba/corba.html#co1 [Accessed July 29, 2010]. Rotem-Gal-Oz, No Date. Fallacies of Distributed Computing Explained. Rgoarchitects. [Online] Available at: http://www.rgoarchitects.com/Files/fallacies.pdf [Accessed July 29, 2010]. Wilkes, J., 1986. Making Distributed Systems Work" position statement. The Wilkes family website. [Online] Available at: http://www.e-wilkes.com/john/papers/1986-EW-wilkes.pdf [Accessed July 29, 2010]. Read More