Distributed Storage Networks allow us to create tightly coupled
remote storage systems. While this is great for disaster recovery of
mission critical systems, it does not lend itself to the construction of
very large scale systems.
The recent advent of service oriented architectures (SOA) has
caused the computing industry to reconsider the fundamental
tenets upon which we build our systems. We are learning that true
scalability comes from the autonomy of smaller systems. In effect,
our subsystems become decomposed into services that provide a
well-defined function for other systems and messaging that allows
communication between these autonomous services. As we
increasingly build our systems according to this paradigm, we will
better define our interfaces and create new discovery methods for
these services.
As multiple services become available, we will be increasingly able
to negotiate for the best terms. Suppose that two services are
available which provide similar but not identical functions. If both
services are "sufficient", which one do we choose? The answer will
depend on many factors. Are there any additional features which
while not necessary are still desirable? Does one service provider
promise to complete the task in a shorter time? Does one service
provider have a better track record than another?
Does all of this sound like the evolution of society?
As we build the first levels of infrastructure needed to achieve
service oriented architectures, we should realize that we are only at
the beginning of a very exciting phase in the development of
computer science.
We're in for an exciting ride.
remote storage systems. While this is great for disaster recovery of
mission critical systems, it does not lend itself to the construction of
very large scale systems.
The recent advent of service oriented architectures (SOA) has
caused the computing industry to reconsider the fundamental
tenets upon which we build our systems. We are learning that true
scalability comes from the autonomy of smaller systems. In effect,
our subsystems become decomposed into services that provide a
well-defined function for other systems and messaging that allows
communication between these autonomous services. As we
increasingly build our systems according to this paradigm, we will
better define our interfaces and create new discovery methods for
these services.
As multiple services become available, we will be increasingly able
to negotiate for the best terms. Suppose that two services are
available which provide similar but not identical functions. If both
services are "sufficient", which one do we choose? The answer will
depend on many factors. Are there any additional features which
while not necessary are still desirable? Does one service provider
promise to complete the task in a shorter time? Does one service
provider have a better track record than another?
Does all of this sound like the evolution of society?
As we build the first levels of infrastructure needed to achieve
service oriented architectures, we should realize that we are only at
the beginning of a very exciting phase in the development of
computer science.
We're in for an exciting ride.
Service Oriented
Architectures
Architectures
Service Oriented
Architectures (SOA) offer a
method of creating
loosely coupled, scalable
computing systems.
They will enable a new
wave of explosive growth
in the way that computers
interoperate with each
other and with mankind.
Architectures (SOA) offer a
method of creating
loosely coupled, scalable
computing systems.
They will enable a new
wave of explosive growth
in the way that computers
interoperate with each
other and with mankind.
Andy Helland
 | |  | |  | |  | |  | |  |