Developing
Complex Systems of Systems Using Cloud and Evaluating their
Architectures Using Simulation and Benchmarking
With
funding from the NSF NCSS I/UCRC and Google App Engine Research
Award, a
team of researchers led by Professor Lawrence Chung
and several student and industrial researchers are helping
understand why complex systems of systems need to be built and
evaluate their architectures using simulation and benchmarking. This research team includes
senior as well as female U.S. citizens as participants and is
motivated by the increasing cost of initial purchase, re-purchase,
and operation of computing equipment that has become unsustainable
and, hence, is becoming an increasingly great burden on the US
economy. This research investigates how to build, evaluate and
select computing resources in a fast and inexpensive way, with a
special emphasis on Cloud Computing as a sustainable and economic
computing paradigm.
The
research objective includes determining if it is possible to
predict:
-
whether an operational system
can migrate to a cloud, while making everyone happy;
-
the performance and
scalability of the system after or even before it is actually built.
The
outcome of the research provides a step towards answering this
question. The research team is investigating “how to make everyone
happy” by using a goal-oriented approach, and how to confirm and
reconfirm whether the architecture of a complex system of systems
will live up to everyone’s expectations in a fair manner by using
simulations.
Technically
speaking, the team captures the stakeholders’ goals, including
cost, performance and scalability (how well the system
will respond with increasing workloads), together with workflows
and reflects them in a good topology of a system of systems. A
system can be a subway monitoring system or a ticketing system, out
of which a more complex system, such as a transportation system, can
be built. Even this transportation system can in turn become a part
of yet another bigger system, such as a system of public facility
systems.
The
space of architectural design for such a system of systems is
usually huge, hence making it practically infeasible to try out each
and every possible design. This research instead explores better
architectural alternatives using simulations – better with respect
to the stakeholders’ goals which often tend to conflict with each
other. It would be a daunting challenge to understand, develop, and
successfully operate a complex system of systems, and this
goal-oriented, simulation-based approach provides a fast way with
little financial and manpower resources to tackle the
challenge.
Benchmarking
is the
process of comparing the metrics of one system to the metrics of
another system, oftentimes industry bests or best practices. For
example, if an organization wants to migrate their system to a cloud
and wants to know if they can get the kind of performance they need
and how much such service can cost, they can consult a similar
system with the best performance characteristics and see how much
cost is involved. Since benchmarking usually involves a reasonably
simple comparison, it can
be even faster and less inexpensive than simulation for the purpose
of prediction. Benchmarking can also help reduce the space of
architectural design to be simulated, since simulation can be
centered around the configuration of the architecture with the best
performance and cost characteristics.
In a nutshell, this research will demonstrate a fast
and inexpensive way of exploring, evaluating, and selecting among
architectural design alternatives as per stakeholder goals. As a special case, it
provides a rational decision support for Cloud Computing, which
seems to be among the most critical technological innovations for
cost savings, especially in these tough economic times. This is also
one of the first proposals for a rational and systematic way of
transitioning from stakeholder goals to the architectural design of
a cloud computing-based system. The work demonstrates the value of
goal-oriented – i.e. a rational - approach to the science of design.
It also demonstrates the value of simulation and benchmarking in
understanding and conquering the complexity of a (cloud-based)
system of systems.