HIGHLY EVOLVED SOFTWARE: SCALE-FREE NETWORKS
The resilience of network architecture is improving as scale-free
systems become more pervasive. In any large organisation today, there is no excuse for network failure.
In the natural world, information is an essentially distributed resource.
This information is not fully contained in central units, rather it is distributed across a large number
of elements that are spread far out in time and space. As a result, natural systems are seldom able to handle
all the information available at any given time. However, despite not having all the information, all the time, natural organisms
and ecosystems have continued to evolve over time.
Although existing computer data bases and networks are built upon highly centralised hierarchies, this is only really effective on a small scale.
As the size of a network increases, centralised networks degenerate, becoming difficult to maintain and scale-free networks easily outperform centralised ones.
Fundamentally, there is no direct exchange of information between computers in centralised networks. In other words, all traffic passes through a server, such as a
Public Switched Telephone Network (PSTN), a strongly centralised communications network. Although this technology has allowed us to place phonecalls through the operator,
and has given us the Internet, other technologies such as IP telephony and P2P networks are just a few examples of breakthroughs by small, highly-skilled companies.
TechIdeas has developed a new functional paradigm whereby self-managing software becomes the critical core of high technology systems.
The main goal of distributed networks is to use a large number of nodes with variable connectivity in a unified form so as to minimise central organisation. There are four main reasons for this:
- to avoid having a single point of failure
- to counterbalance resource limitations
- it permits a more efficient use of distributed information and resources
- it increases the overall performance and stability of the system.