Advertisment

Networks of the future

author-image
DQW Bureau
New Update

Convergence, it appears, is going to be the core inspiration for networks of the near future. The question that arises is whether the present day, e-commerce based client-server architectural platform for enterprises and the web is capable of addressing some of the potential issues that seem to creep up. It is therefore imperative to acknowledge some of the fundamental characteristics of the present day network based systems, so that we understand and appreciate how different futuristic convergence based networks are going to be.

Advertisment

The unique characteristics of current computing and network architectures 

Centralized governance: Perhaps the most distinguishing characteristic of web-based network architectures that distinguishes them from previous three-tier client-server architectures is that the clients have become very thin and light.

The complex business processes are tirelessly handled by a central administration system and only the results are spelt out to the clients-thus rendering the client terminals almost dumb. The point of interest here is the presence of a central governing authority (called the `server') - that takes the responsibility of handling client requests that keep pouring in throughout the day. Web servers (or the Internet servers) that were primarily developed to handle Html requests and responses, had a lot of difficulty in coping with business processing or transaction requirements of the client's requests.

Advertisment

This initiated the need for an entirely new breed of servers that are now called the 'Application Servers', to do business processing in a different tier.

Almost all back end 'resources' (for e.g. database), as well as the business process modules, have to make themselves fully available for the server that might refer to them as and when it is necessary. Thus, a server must be fully aware of all its resources and modules so as to decide whether it has enough resources and inputs to handle a client's request or not.

It is interesting to note that the clients are not aware of any of these back end resources. The back end remains as a `black box' for the clients. While this is an advantage from the point of security, it is not so from the point of convergence.

Advertisment

This obviously makes the server more and more complex, bulky and difficult to handle. This also implies that a particular server is tied (or dedicated) to a specific set of business tasks. If a new business process has to be added or a new resource has to be plugged away, these proceedings have to be informed to the server so that it becomes aware of the structural changes. This introduces a certain degree of inflexibility and makes incessant changes difficult to perform.

Moreover, when several business modules that make up a system are piled up in one place, it soon becomes an administrator's nightmare-rendering it more and more complex. Reaching a specific service becomes a complex task and unnecessary time and computing power is wasted simply to identify where the module that is capable of performing the specific task requested is located.

The 'poor' thin client: The third distinguished characteristic of this architecture is that the computing power that is available at the client's end is almost fully wasted. Nowadays, it is not uncommon to see a robust PIII machine used only for checking emails and browsing the net! The client is capable of performing a good number of business tasks it has actually requested the server to do- but the problem is that the business algorithms and resources are only available to the server and are not available for the client. Thus, it has to wait for the server to allot its time for completing the business process requested, while it can take care of the same by itself when it is endowed with sufficient resources. Thus, thin client architecture results in wastage of immense computing power available at the client's end and at the same time, necessitates the need for more and more power at the server's end.

Advertisment

Everything is a computer: The fourth major and final point is something that is related to the current trend in convergence. With almost all telecommunication and entertainment devices being increasingly controlled by microprocessors, there is a need for any system to make its services available to a whole lot of strikingly heterogeneous clients that may require its service.

Convergent networks: Convergence refers to that methodology by which a variety of diversified technologies and equipment come together to offer a set of services to the consumers. The simplest example would be a cable modem that enables a television cable to carry data from the worldwide Web-to be viewed on a television-rather than a computer. In this case, a well-diversified variety of equipment and technologies- telecommunication, computer processing, data transfer and web-all come together, to offer a single service.

With almost all consumer electronic devices employing what are called 'embedded technologies' it is important to think of a computing world that suddenly breaths to life outside a PC, for e.g. Internet screenphones, wireless smartphones, intelligent refrigerators, etc. It is obvious that we are slowly moving towards a world that is entirely different from the present day computing and networks - a world that is closely interconnected by a variety of consumer electronic devices--a world in which all technologies would converge together to provide a set of well defined services to the user.

Advertisment

Fundamental features of complex and heterogeneous networks of the future:

  • Multiplicity of clients: The variety of clients the network has to handle would be too many. Moreover, the variety of interactions that can happen across them would also be much more complex and unpredictable. It is impossible to make a single server aware of all these varieties of client requests and interactions. It is impossible to have a single centralized controlling authority, given the number and varieties of business tasks and client requests. This completely violates the very basis of client-server computing, where the services are available in a central server and all clients come in with their requests! We are talking about a world where there is no specific `server' that is dedicated to 'serve' client requests! This then takes us to the next level of complexity involved in this business:

  • Advertisment
  • Distributed servers: When there is no central authority that is dedicated to offering a set of services, then it is automatically implied that not only the clients, but the services too, are fully distributed and not available in one place! This becomes more complex so we think of an integrated service, which in turn is dependent upon certain other services, to complete its business task! It is true that distributed computing is nothing new and we already have many applications and resources like databases on the web that are distributed. But, what is strikingly different is the absence of a central authority that coordinates the transactions between these distributed services.

  • Look-up services: When the services are distributed in a network, there is obviously a need for a place where the information about all the services that are available on the network is available. For example, in the Internet, we almost fully rely upon search engines to locate those sites. These kinds of look-up services are already made use of by the distributed development environments of the present day networks (they are called Naming and Directory services- IBM's LDAP being a good example of the same). All we need to do is to adopt them for the heterogeneous convergent network we would eventually be constructing.

  • Plug and play: In this fluid network of the near future that we are perceiving, a system or module that is capable of performing a specific business task must be able to `Plug in' seamlessly to the existing network and its services should be easily available to the clients.

  • Advertisment
  • Client-service contracts: When a particular client is in need of a service or set of services, it comes into contact with the service providers scattered across the network. This client may, in turn, be a service provider also. Thus, we see that a set of clients and services join together in a network for completing their respective business tasks. This grouping is very dynamic, in the sense that a new set of clients and services keep adding onto the network on one side and old clients and services keep detaching themselves from the network on the other side. The relationship that exists between a service and client is purely temporary. At a later point of time, the client may even satisfy its needs with a new service provider, unless it has specific reasons to look out for its old pal. This dynamism is in striking contrast to the present day Internet work, where the service provider, at least, is not so dynamic and almost all kinds of client requests are obviously from the browsers.

  • Industry initiatives

    Almost all the technology majors are fully aware of the implications of convergent networks and are gearing up their efforts to offer reliable infrastructure and solutions for the same. One of the most noteworthy efforts in this direction is by the Internet work technology leader-Sun Microsystems-which has released a dynamic network architecture and system initiative called JINI, very specifically aimed at addressing most of the potential issues we have been discussing so far. While the JINI initiative is not free from Sun's own fundamental technological throughputs like Java Virtual Machine as the base development environment and concepts like RMI (Remote Method Invocation), it is certainly most notable in the goals it tries to attain by means of
    JINI.

    If more and more companies decide to invest heavily on their e-business infrastructure, then we are left to think that they may have a larger shelf life than expected and that the convergent networks will be limited in their scope. They will be useful in developing smaller groups of 'federations' inside corporate environments and homes and in turn, get themselves connected to the rest of the world only through a proven Internet gateway.

    Ultimately, we need to wait and see whether these two different approaches to networking remain as two different worlds that decide to co-exist or whether they succeed in relocating each other in the world marketplace!

    S Gokul is an Internet and a multimedia consultant with Cybervalley Systems.
    Source: www.voicendata.com

    Advertisment