Raytheon recently developed and implemented a Near Real Time (NRT) data processing subsystem for Earth Observing
System (EOS) Microwave Limb Sounder (MLS3) instrument on NASA Aura spacecraft. The NRT can be viewed as a
customized Science Information Processing System (SIPS) where the measurements and information provided by the
instrument are expeditiously processed, packaged, and delivered. The purpose of the MLS NRT is to process Level 0
data up through Level 2, and distribute standard data products to the customer within 3-5 hours of the first set of data
The Science Data Processing System (SDPS) application, described herein as an example, has had a long development
cycle. The SDPS application incorporates IDL, C++, and Perl programming languages, with significant use of an Oracle
Relational Database Management System (RDBMS). The time involved from initial design, 1996, to operational
deployment, on July 15th, 2004, with the launch of the Aura spacecraft, spans several releases of the Oracle RDBMS.
New database versions usher in new features and performance enhancements, sometimes requiring modifications to
custom application code to take full advantage of improvements in technology. For a project with an aggressive release
schedule, major redesigns of the custom code could jeopardize the successful completion of mission criteria.
Over time, higher throughput requirements and hardware improvements in the application computing infrastructure
revealed database performance bottlenecks due to increased scanning frequency of ever-growing tables and indexes. The
Distributed Schema (DS) database redesign successfully addressed the database scalability and performance limitations,
with only minor SIPS application changes and no changes to the TES SDPS application.