Abstract
This paper concerns the comparison between centralized and distributed database storage and management systems. The prime objective of the study is to explore the limitations of centralized databases and databank management systems and establish the how the limitations were overcome in distributed databases and database management systems. The challenges and the solutions are based on today’s business requirements and dynamic needs of organizations. The forcers which drove the shift from centralized to distributed platforms are explored and include the emergence of client-server architecture, Internet computing, local and wide area networks and the World-Wide-Web. The advancement of major structures in prominent database and database management systems in previous decades made the upgrade to distributed platforms inevitable. Traditionally, organization’s data resources were centralized due to the origin of database management within the mainframe environment. The research will also explore the potentials held by distributed databases in providing organizations with competitive edges.
Introduction
The utilization of information systems in organizations usually covers eight major segments. They include information system planning, organizational structures and human capital, data management, computing and data management architecture, information system development, system acquisition, and training and support. Data management component is the greatest concern for organizations owing to the advancement in technology furthermore, advancement. An association with geologically scattered areas may choose to store its information in a focal database server situated at the home office or appropriate information on nearby servers or both. A decision of both of the alternatives is reliant on the authoritative structure of the association among different components. Embracing an incorporated way to deal with information stockpiling and administration can achieve productivity yet obliges some extreme imperatives to be succeed. A decentralized choice, on the other hand, can help spread computing in the organization but is often wasteful. Some organizations may adopt a combination of the two depending on their needs and operational techniques (Tripathi, 2011).
Large commercial databases exist in two different topologies: centralized and distributed. A consolidated database is based in one physical location for access by many users from different segments through the network connection. An instance of such a database is a bank’s database and database management system. A distributed database is available in many locations where users interact with a locally. It is characterized as a solitary consistent database spread physically across computers in many locations connected by a network. For instance, Google is an international company using Big-Table distributed DBMS for searching trends by users in particular locations of the world. The conveyed database is a genuine database that is directed midway as a hierarchical asset while giving nearby adaptability and customization. Clients are associated through a system which allows them to share data such that a user at location A can access data at location B. The site of a distributed database may be large, for example, the entire US or small such as a campus.
The evolution from centralized to distributed databases is facilitated by business and technological forces. Preceding the well-known acknowledgment of DDBMS, enterprises depended on brought together databases to serve the organized data necessities. The structured information was presented as regularly issued formal reports in standard formats. The use of centralized databases required corporates to store their data in single principal sites, preferably supercomputer workstations. Data access was through serially connected dumb terminals. The centralized approach served well to meet the structured information requirements, be that as it may, it felt shy of velocity in moving occasions and quickly accessing them. In the 1980’s, a series of social and technological innovations impacted the development and use of database systems. For example, business operations became more decentralized globally while competition surged. The result was a client demand and market needs that favored a decentralized management style.
Furthermore, rapid technological change, especially the low-cost mainframe, facilitated the development of distributed architectures. The issues revolving around handling of large, data-centric applications over file-sharing networks lead to the development of client/server architectures in the 1980s. Prior to this, the wide acceptance of the personal computer with its native computing and graphical user interface made it possible to increase the complexity of applications and expand network systems to the second major type of system architecture that is file sharing. In the client/server architecture, the file server is replaced by the database server which in addition to transmitting and saving files to the customers, gets and executes demands for information, yielding just the outcome sets to the customer. The customer/server gave an inquiry reaction instead of a complete file transfer format and this facilitated the development of application that allowed multiple users to update data via graphical user interface front ends connected to single shared databases. The two most common methods of communication between the client and the server are structured query language and remote procedure calls.
A large number of applications based on DBMS and the need to protect investments in centralized DBMS software became more pressing and made the notion of data sharing attractive. During the 1990s, these factors were more pressing than before and was influenced by the growing acceptance of Internet and www platforms as the platform for data access and distribution. The growth of internet and explosion of web pages containing real-time information dramatically increased the demand for dynamic business web pages that relied on dispersed databank management.
Decentralized database management systems grew with client-server architectures. The potential for further growth is enormous and tied to client-server. Experience with client-server architectures has indicated that the complexity and expense of these approaches can be overwhelming. As a result, some organizations has decided to go back to the direction of brought together databases. Experts have reacted to the test by creating arrangements utilizing servers taking into account centralized server or minicomputer stages and slight customers which in many occasions is keep running on web programs. The construction modeling is bended out of customer server and is alluded as web processing. The advocates guarantee that the disentanglement of the distributed components of the architecture and movement of data to one professionally-managed location delivers reliability and reduces the cost of implementation. One of the novel arguments for a client-server was the ability to replace character-based terminals with GUI-based workstations that are very flexible and easy to use. Internet computing takes up the benefits of a central data store and GUI-based workstations and may generate stiff competition for widely distributed DBMS systems.
The immaturity of the client-server architecture in the past decade or two limited the advancement of distributed DBMS. Many industry professionals felt that until the three-tier architecture becomes the norm, the benefits of client-server can't be figured it out. In like manner, a dispersed database obliges four noteworthy facilitators that include: equipment freedom, working framework autonomy, system freedom and database autonomy. In any case, today's DMBS don't meet every one of these benchmarks and regardless of the possibility that the circumstance is enhancing, the cross—merchant integration is still an issue particular for legacy systems that do not implement new standards.
The infusion of PCs and LAN in the workplace took center stage in the 80s and contributed to the development of the decentralized database systems. PCs started controlling organization's corporate workstation in large numbers, and the natural extension of these machines being on many desktops meant that there was connection capability using local networks and servers. Office fileservers gave little organizations decentralized server power. The thought and foundation of corporate reflected an increasingly decentralized bias-motivated in many instances by end-users who began to understand and experiment the latent of decentralized computing.
Business leaders appreciated the need for information as an organizational resource. With compacted storages, an organizations information is maintained and managed by a limited intellectual personnel based at a single location. Two major factors contributed to the rejection of centralized databases. One, the normal perceptions of people not to disseminate information and two, and the emergence of PC-based DBMS systems that were dominant as to control multiple concurrent instances. With the implements, organizational sections and working groups could build their databases, fight control of information from the administrators of the organizations central databases and attain their deep-lying belief not to disseminate information (Subramanian, 2007).
In the early years of a concentrated database, experts endorsed an one-size fit all idea to DBMS applications. On the other hand, as databases developed, it got to be clear that DBMS should be particular to bolster the expanding and unique business exercises. One of the noteworthy movements happened in the server farms running centralized computer PCs. Centralized servers utilized social databases improved for exchange speed however in times of report and question preparing, low execution were noted. As of right now, there were diverse prerequisites for online exchange handling and online investigative preparing frameworks. While a few associations reacted to execution challenges by limiting OLAP to night and off-crest periods, it confined the utilization of OLAP as it didn't upgrade the game changer it was intended to offer. For others, it was unrealistic since they worked on a round the clock premise. (Guynes, 2011)
Fragmentation
Fragmentation in circulated databases is an idea that needs much comprehension. Fragmentation is accomplished by dividing the unified database into little sections and scattering them to diverse areas. The appropriation is accomplished by means of even or vertical parceling. In vertical discontinuity, no information is put away needlessly aside from the essential keys. Embracing the relational model, even fragmentation is refined by isolating lines and vertical discontinuity is accomplished by isolating sections. Information is divided by area of the association that uses or changes it every now and again. For example, an organization may utilize an office code to question the division that handles a record and focus the area where information ought to dwell. It is excessive to ship huge amounts of information over a correspondences system or handle substantial volumes of information from a remote site. In this way, the idea of living information where it is utilized most as often as possible is helpful as a part of distributed database frameworks. In the same route, reliance on information correspondence may posture dangers and henceforth, information replication or duplication is used in decentralized frameworks. The key standard of divided distribution is that just a solitary duplicate of information exist in the database with numerous proprietorship and capacities to upgrade. In spite of the fact that divided database frameworks are more perplexing than concentrated frameworks, they are less complex than duplicated frameworks. The test of a solitary point of disappointment is decreased, and network use is lower in divided frameworks than in brought together frameworks.
According to Silberschatz, Korth and Sudarshan, (2009) replicated databases are beneficial than traditional means of failure recovery associated with centralized database systems. Replication provides fault-tolerance beyond what can be achieved using a redundant array of inexpensive disks. Replication makes the database to be physically accessible in different areas on the system, and the chance that disappointments will bring about loss of administration is essentially lessened. In a distributed database, there are two options that facilitated the failure recovery: warm standby and hot standby. Warm standby denotes asynchronous duplication to maintain the standby server in a manner nearly consistent with the primary server. A small application loss is appreciated in warm standby due to the lag between transactions committed on the primary server and replication to the standby server in cases of a primary server failure and switchover to the standby server. Hot standby uses synchronous replication to keep the standby server in a state always consistent with the primary server. From an availability standpoint, this option is preferred for database systems but its cost of implementation is slightly high (Kroenke, 2011)
The database management system industry is highly competitive with no single player dominating the industry. Most of the DBMS offerings from major vendors have the distributed components on them. The major differences between the products are the technical aspects of how data distribution is achieved. The database management system market is so competitive with major players such as IBM, Oracle, and Microsoft engaging fiercely. Overall, DDBMS have done a remarkable job in meeting the necessities of organizations. Effective use in system associations has made it simple to get to information topographically scattered. Likewise, system loads associated with replication have decreased in recent times. The latest trend is replication to small client databases using dial-up connections providing organizations with powerful tools for meeting their requirements and supporting mobile users. With cautious design and scrutiny of business requirements, current DDBMS can support geographically dispersed business operations. This is preference when contrasted with brought together databases.
DDBMS have the capability to provide information managers with a means of centrally controlling and utilizing information scattered by server proliferation. There are effective implements that are suited for handling heterogeneous server platforms and replication that is flexible for legacy applications. Because the distributed phenomenon works best with centralized planning and decentralized operations, database designers and managers have an edge in controlling a firm’s information flow. The dispersed database can also permit mergers and acquisition for organizations looking forward to such plans. Previously, consolidated databases could not facilitate merging of information from multiple platforms.
Conclusion
Recording management system developers have made great improvements in handling of high loads of simultaneous OLTP and OLAP data. Recent advances in technology such as the improved use of processor hardware, multithreading, and client-server architecture have facilitated the development of decentralized database systems that are efficient and economical as compared to centralized systems. The replication features of today’s DDBMS facilitate speedy access, error tolerance and security of data in OLAP and OLTP servers. Previously, these features could not be solely achieved in centralized systems.
References
Abraham Silberschatz, H. K. (2009). Database System Concepts 6th edition. McGraw Hill.
Anand K. Tripathi, M. T. (2011). A Framework of Distributed Database Management Systems in the Modern Organization and the Uncertainties removal . International Journal of Advanced Research in Computer Science , Vol 2 No. 5.
Ashok Subramanian, M. L. (2007). Managing Client-Server implementations: Today's technology, yesterdays lessons . Journal of Information technology.
Chauhan, M. A. (2011, July). Migrating service-oriented system to cloud computing: An experience report. In Cloud Computing (CLOUD) . 2011 IEEE International Conference on Cloud Computing, (pp. pp. 404-411).
Guynes, C. W. (2011). Revisiting Client/Server Computing. Journal of Business and Economics.
Heeks, R. (2009). Information Systems for Public Sector Management: Centralized vs. Decentralized. University of Manchester.
Kroenke David, A. D. (2011). Database Processing: Fundamentals, Design and Implementation. Perason.
Noel, H. (2013). Distribiuted Databases. Freeman.