Improving productivity and efficiency through a multistage implementation

Financial services firms can take an
existing inefficient infrastructure for
risk management and compliance
and gradually grow it into an integrated,
highly efficient grid system.
As shown in Figure 1, an existing
infrastructure may comprise stove
pipes of legacy applications disparate
islands of applications, tools
and compute and storage resources
with little to no communication among
them. A firm can start by enabling
one application a simulation application
for credit risk modeling, for
example to run faster by using grid
middleware to virtualize the compute
and storage resources supporting
that application.
The firm can extend the same solution
to another application, for example,
a simulation application used to
model market risk. Compute and storage
resources for both simulation
applications are virtualized by
extending the layer of grid middleware;
thus both applications can
share processing power, networked
storage and centralized scheduling.
Resiliency is achieved at the application
level through failover built into the
DataSynapse GridServer. If failure
occurs or the need to prioritize particular
analyses arises, one application
can pull unutilized resources that are
supporting the other application. This
process also facilitates communication
and collaboration across functional
areas and applications to provide
a better view of enterprise risk
exposure.
Alternatively, a firm can modernize by
grid-enabling a particular decision
engine. A decision engine, such as
one developed with Fair Isaac’s tools,
can deliver the agility of business
rules and the power of predictive analytic
models while leveraging the
power of the grid to execute decisions
in record time. This approach
guarantees that only the computeintensive
components are gridenabled
while simultaneously migrating
these components to technology
specifically designed for decision
components.
Over time, all applications can
become completely grid-enabled or
can share a common set of gridenabled
decision engines. All compute
and data resources become one
large resource pool for all the applications,
increasing the average utilization
rate of compute resources
from 2 to 50 percent in a heterogeneous
architecture to over 90 percent
in a grid architecture .
Based on priorities and rules,
DataSynapse GridServer automatically
matches application requests
with available resources in the distributed
infrastructure. This real-time brokering
of requests with available
resources enables applications to be
immediately serviced, driving greater
throughput. Application workloads
can be serviced in task units of milliseconds,
thus allowing applications
with run times in seconds to execute
in a mere fraction of a second. This
run-time reduction is crucial as banks
move from online to real-time processing,
which is required for functions
such as credit decisions made
at the point of trade execution.
Additionally, the run time of applications
that require hours to process,
such as end-of-day process and loss
reports on a credit portfolio, can be
reduced to minutes by leveraging this
throughput and resource allocation
strategy.