10/31/2018

Open Architecture Computing Environment
3rd Workshop on The ACE ORB (TAO)

Charlie FudgeTraci McDonald
NavalSurfaceWarfareCenter Dahlgren Division
{fudgeCL|McDonaldTA1}@nswc.navy.mil

The motivation for the Open Architecture (OA) effort is to design and build Naval Weapon Systems that not only continue to meet changing performance requirements well into the 21st century, but also may be rapidly and affordably maintained, and refreshed throughout their life-cycle. Design goals also include enhanced Human Systems Integration (HSI) and optimized manning.The current issues that must be addressed by OA are as follows:

  • Architectural obsolescence
  • Frequent changes and upgrades to the combat systems
  • Applications tightly coupled to the underlying infrastructure
  • Non standards based applications

These issues have contributed to high estimated costs for system development (including re-architecting each new system), and computer program maintenance, as well as equipment and support software technology refreshes. This high cost has now become a barrier to continued system development and fleet support. The evolution of high performance COTS, however, combined with continued growth of weapon system and combat system requirements, provides an opportunity to design an architecture more capable of exploiting new technologies than the federated legacy architecture that has served the Navy for well over two decades. The need for evolution toward an open architecture is motivated by both performance and supportability considerations. Commensurate with this dual set of motivating factors, the goals of the OA are as follows:

  • Combat system, weapon system, command support system and HM&E capabilities that continue to pace the threat
  • System design that fosters affordable development and life-cycle maintenance
  • System design that reduces upgrade cycle time and time-to-deployment for new features
  • Architecture that is technology refreshable despite rapid COTS obsolescence
  • Improvements in NWS Human Systems Integration

These goals will be met, in part, by careful architecture and design and in part through the use of open systems principles and standards.

Standardized middleware, such as the Common Object Request Broker (CORBA) is projected to be a major infrastructure component in the Open Architecture Computing Environment (OACE), and will be instrumental in managing change by isolating the applications from the underlying computing technology.

The High Performance Distributed Computing (HiPer-D) Program at the Naval Surface Warfare Center Dahlgren Division (NSWCDD) was established to explore distributed computing concepts, technologies, and architectures, and to evaluate their applicability to complex Navy shipboard systems, such as the Aegis Weapons System (AWS). Directly relating to the OA initiative,HiPer-D has performed system-scale prototyping directed toward the new open architectural principles being pioneered in the commercial world.This is accomplished through evaluation of concepts, architectures, and technologies that support the HiPer-D reference architecture in the context of a prototype system-scale testbed. Specific products are also assessed through stand-alone evaluations and benchmarks. Currently, the tactical applications that comprise the HiPer-D test bed are using CORBA, including multiple real-time ORBs . In addition to the Hiper-D test bed, CORBA products are installed in the EDM(0) lab. EDM(0) is an instantiation of the OACE that will be used for integration and testing of OA applications.

The Navy envisions that CORBA will play a significant role in OA. In anticipation, we are examining multiple aspects of using CORBA. We are looking both at current capabilities as well as future possible directions through evaluations and benchmarking, interactions with the research and vendor communities, use of CORBA within the prototype testbed and EDM(0), and participation in the Object Management Group.

This presentation will discuss the results of a standalone benchmarking effort involving two CORBA products, TAO and ORBexpress. The purpose of the effort was two fold; 1), determine how much effort was involved in porting existing CORBA benchmark applications from one OS to another with a different ORB, and getting those ORBs to interoperate; 2), determinehow the performance scaled for different ORB (TAO and ORBexpress), OS (Solaris and Lynx), and hardware (x86 and Sparc) configurations from the ideal (same ORB, same OS, and same hardware). Looking ahead to an OA implementation involving CORBA one can envision multiple configurations of ORBs, OSs and hardware involved in intra and inter subsystem communication.This project provided insight and data on two key implementation issues, product interoperability and product interchangeability. Specifically, the presentation will contain introductory material stating the motivation for the project and illustrating the equipment and software configuration. Data charts will be showndepicting latency and throughput measurements for various data types and structures for TAO and ORBexpress in a number of different configurations. The presentation will conclude with several summary slides stating conclusions related to the goals of the project.