Professional Joomla Templates Free

Distributed systems have evolved considerably in recent years. Depending on the scalability level required, several solutions can be found to develop distributed applications. If the number of users is relatively low, there are several centralized client-server models, with a rather simple subjacent architectural complexity, which provide an acceptable performance.

As soon as a distributed application needs to grow, its set of target users probably does so as well. It is at this point that centralized solutions start showing performance problems because of the well known bottleneck effect: that is to say, the central server is unable to efficiently cope with all of the users‘ needs. As a consequence of these scalability needs, the current trend is towards decentralization, thus minimizing this bottleneck effect by adding more servers that process logic. Therefore, not so many clients overwhelm the single central server.

These approaches are adequate for applications which do not need to be worldwide scalable because the number of potential users they are used by is relatively enclosed, and not excessively big. However, when worldwide scalability is required, decentralization is one of the best options, but it also creates many problems which have to be dealt with. Some of these problems include how to provide fault tolerance, how to deal with constant node joins and leaves, and many others.

This thesis proposes a set of middleware elements that help to develop distributed applications which are worldwide scalable, dynamic and fault tolerant. We have chosen a totally decentralized paradigm which guarantees the scalability of worldwide accessible applications. This middleware is built on three basic pillars. Firstly, we need an efficient message routing layer; secondly, we need a layer which provides application-level multicast services, so that one-to-many communication is efficient. Finally, a decentralized persistence service is required to store and restore data.

Using these three primitives, we build a wide-area middleware platform for the development of distributed applications, based on a remote object layer and a distributed component layer. The functionalities of these tiers‘ are as generic as possible so they do not depend on the underlying layers. Therefore, several underlying decentralized paradigms can be theoretically used without any variation in the middleware layer. Consequently, developers dispose of a higher level of abstraction when building wide-area distributed applications, since they can make good use of the services provided by the middleware layer.

In order to materialize this thesis, we implemented the proposed middleware framework on top of a relatively new paradigm of peer-to-peer networks, called structured peer-to-peer networks. These networks have interesting properties which make them highly suitable for the development of new distributed services.

The main contributions of our wide-area middleware proposal include the design and definition of a brand new set of invocation abstractions for our remote object model; a remote distributed interception algorithm; and a decentralized object location service. We have also defined two new load balancing algorithms based on the primitives of the remote object layer. As far as the reusable distributed component layer is concerned, our main contribution is the adoption of a lightweight container model which allows decentralized deployment and the activation of reusable components.

We also provide proof of the concept of our architecture with the implementation of SNAP: a decentralized, fault tolerant and scalable Java Platform, Enterprise Edition (J2EE) [46] web application deployment platform. The idea is to facilitate already existing J2EE applications in a worldwide interconnected network, in the easiest possible way, and by changing as little code as possible.

With the vision that we provide, we believe that developers will be able to produce newly decentralized applications by investing a minimal amount of learning time in a technology which surely still has much to show us.