The Need
The retail chain needs to process sales
results on a per-product and per-time-of-day basis.
Results need to be available fifteen minutes after
every business hour to enable middle-of-the-day changes to
in-store marketing strategy, pricing, or product
placement.
The Challenge
Hourly sales results need to
be analyzed in less than fifteen minutes.
Multiprocessing is essential to provide the
speed and reduce the output delay. However,
dedicating CPUs to multiprocessing would leave them
dramatically under utilized for most of each hour and
completely unutilized overnight.
The ReFrame Active
Service Grid Solution
Applications developed for
the ReFrame Application Service Grid can scale dynamically.
ReFrame's message-based architecture, frame-based
operation and zero-configuration networking make
distributing and scaling simple. Dynamic
Scaling adds processing power when needed and leaves
the CPUs available otherwise.
ReFrame enables development of an algorithm that uses
CPU load and other parameters to define when to hand
off frames to other CPUs. This same algorithm
can return those CPUs to the availability pool when the
work load no longer requires them.
ReFrame's ability to enable
Dynamic Scaling allows multiple applications to share
a pool of available CPU power. Participating
applications run faster, results that drive the
business come sooner. CPU
resources are shared which raises utilization levels
and lowers both capital and operating costs.
Accommodate Change -
Fast, Easy, Safe
As this retail chain both
adds stores and increases same store sales,
processing power can scale by adding computers to
the dynamically scaled ReFrame Application Service
Grid.
As new information needs arise, the applications
written for the ReFrame Application Service Grid can
be modified and new ones added.
ReFrame enables you to quickly, easily, and safely
accommodate the changing size and shape of your
business.
|