The Center for Distributed Object Computing Research Synopsis

Transcription

1 The Center for Distributed Object Computing Director, Center for Distributed Object Computing Research Synopsis David L. Levine Computer Science Dept. Sponsors NSF, DARPA, Bellcore/Telcordia, BBN, Boeing, CDI/GDIS, Comverse, Hughes, Kodak, Lockheed, Lucent, Microsoft, Motorola, Nokia, Nortel, OCI, OTI, Raytheon, SAIC, Siemens SCR, Siemens MED, Siemens ZT, Sprint, USENIX 24 November 1999 Motivation: the Communication Software Crisis Symptoms Communication hardware gets smaller, faster, cheaper Communication software gets larger, slower, more expensive Culprits Inherent and accidental complexity Solution Approach Standard communication middleware 1 Research Focus Members Patterns, Pattern Languages, and Frameworks Mission Critical and Embedded Systems Configurable Communication Systems High Performance Real-Time CORBA The ACE Orb () Distributed Simulation Adaptive Communication Environment (ACE) David L. Levine, Director Fred Kuhns, Associate Director Douglas C. Schmidt, Former Director Full-time staff: C. Gill, C. O Ryan, J. Parsons, I. Pyarali, N. Wang PhD students: C. Gill, J. Hu, C. O Ryan, O. Othman, I. Pyarali, N. Wang Masters students: L. Baker, D. Brunsch, P. Gore, V. Kachroo, Y. Krishnamurthy, B. Natarajan, K. Parameswaran, J. Parsons, M. Spivak Undergrads: K. Pathayapura 2 3

2 Sponsors Problem: Lack of QoS-enabled Middleware Bellcore/Telcordia BBN Boeing CDI/GDIS Comverse DARPA Hughes Kodak Lockheed Martin Lucent Microsoft Motorola Nokia Nortel NSF OCI OTI Raytheon SAIC Siemens Sprint USENIX Many applications require QoS guarantees e.g., avionics, telecom, WWW, medical, high-energy physics Building these applications manually is hard Existing middleware doesn t support QoS effectively e.g., CORBA, DCOM, DCE, Java Solutions must be integrated horizontally & vertically 4 5 Candidate Solution: CORBA Goals of CORBA Simplify distribution by automating Object location & activation Parameter marshaling Demultiplexing Error handling Provide foundation for higher-level services 6 Washington Un Caveat: Requirements/Limitations of CORBA for QoS-enabled Systems Limitations Requirements Lack of QoS specifications Lack of QoS enforcement Location transparency Performance transparency Predictability transparency Lack of real-time programming features Lack of performance optimizations Reliability tranparency

3 Problem: Optimizing Complex Software JSAC-99.ps.gz Common! Problems Optimizing complex software is hard Small mistakes can be costly Solution Approach! (Iterative) Pinpoint overhead via white-box metrics e.g., Quantify and VMEtro Apply patterns and framework components Revalidate via white-box and black-box metrics Solution 1: Patterns and Framework Components CLIENT CONNECTOR OS KERNEL COMPONENT CONFIGURATOR ABSTRACT FACTORY STRATEGY THREAD-SPECIFIC STORAGE REACTOR WRAPPER FACADES SERVANT LEADER/ FOLLOWERS ACCEPTOR OS KERNEL Definitions Pattern A solution to a problem in a context Framework A semi-complete application built with components Components Self-contained, pluggable ADTs 8 9 Washington Un Solution 2: ORB Optimization Principle Patterns Definition Optimization principle patterns document rules for avoiding common design and implementation problems that can degrade the efficiency, scalability, and predictability of complex systems Key Principle Patterns Used in # Principle Pattern 1 Optimize for the common case 2 Remove gratuitous waste 3 Replace inefficient general-purpose functions with efficient special-purpose ones 4 Shift computation in time, e.g., precompute 5 Store redundant state to speed-up expensive operations 6 Pass hints between layers and components 7 Don t be tied to reference implementations/models 8 Use efficient/predictable data structures Patterns for Communication Middleware Observation Failures rarely result from unknown scientific principles, but from failing to apply proven engineering practices and patterns Benefits of Patterns Facilitate design reuse Preserve crucial design information Guide design choices 11

4 The ADAPTIVE Communication Environment (ACE) ACE Overview Concurrent OO networking framework Available for C++ and Java Ported to POSIX, Win32, VxWorks, Chorus, PharLap TNT, et al. 12 Washington Un The ACE ORB () and Its CORBA Object Services ACE and Statistics Over 30 person-years of effort > ACE 200,000 LOC > 125,000 LOC IDL > compiler 100,000 LOC CORBA Object > Services 150,000 LOC Ported to POSIX, Win32, VxWorks, et al. Large user community Currently used by dozens of companies Bellcore, Boeing, Ericsson, Kodak, Lockheed, Lucent, Motorola, Nokia, Nortel, Raytheon, SAIC, Siemens, etc. Supported commercially! ACE Real-time Optimizations in 15

5 ' ' & & $ $ % % New Features and Optimizations New Features Real-time CORBA Minimum CORBA CORBA Messaging Fault Tolerance URL schmidt/status.html Integrating with ATM I/O Subsystem schmidt/rio.ps.gz Features Vertical integration of QoS through ORB, OS, and ATM network Real-time I/O enhancements to Solaris kernel Provides rate-based QoS end-to-end Leverages APIC features for cell pacing and zero-copy buffering struct RT_Info { wc_exec_time_; period_; criticality_; importance_; dependencies_; }; 1. SPECIFY RT_OPERATION CHARACTERISTICS AND DEPENDENCIES 8. CONFIGURE QUEUES BASED ON DISPATCHING QUEUE CONFIGURATION 10. DYNAMIC QUEUES ASSIGN DYNAMIC PORTIONS OF DISPATCHING SUBPRIORITY (AND POSSIBLY Strategized Scheduling Framework (SCHEDULER'S INPUT INTERFACE) 2. POPULATE RT Operation RT_INFO REPOSITORY RT Operation RT Operation OBJECT ADAPTER ORB CORE I/O SUBSYSTEM DISPATCHING PRIORITY) ORB ENDSYSTEM OFF-LINE STRATEGIZED SCHEDULER RT_INFO REPOSITORY SCHEDULING STRATEGY (SCHEDULER'S OUTPUT INTERFACE) OFF-LINE 3. ASSIGN STATIC PRIORITY AND STATIC SUBPRIORITY 4. MAP STATIC PRIORITY, DYNAMIC SUBPRIORITY, AND STATIC SUBPRIORITY INTO DISPATCHING PRIORITY AND DISPATCHING SUBPRIORITY 5. ASSESS SCHEDULABILITY 6. ASSIGN DISPATCHING QUEUE CONFIGURATION RUN-TIME SCHEDULER RT_INFO REPOSITORY ON-LINE 7. SUPPLY DISPATCHING QUEUE CONFIGURATION TO THE ORB 9. SUPPLY STATIC PORTIONS OF DISPATCHING PRIORITY AND DISPATCHING SUBPRIORITY TO THE ORB 18 Washington Un Use-cases for ACE and Domains Electronic medical imaging Network management Wireless personal communication systems (PCS) Real-time avionics mission comuting Multimedia services Distributed interactive simulation

6 Applying ACE and to Medical Imaging Large volume of Blob data e.g., 10to40Mbps Lossy compression isn t viable Prioritization of requests URLs schmidt/coots- 96.ps.gz schmidt/av.ps.gz schmidt/nmvc.html Applying ACE to Network Management Low latency Multi-platform Family of related services URLs schmidt/dsej- 94.ps.gz schmidt/ecoop- 95.ps.gz Applying ACE to Global PCS Long latency satellite links High reliability Prioritization URL schmidt/tapos- 95.ps.gz Applying to Real-time Avionics Real-time periodic processing Complex dependencies Very low latency URL levine/research/jsac- 98.ps.gz 22 23

7 HIGH - BANDWIDTH VIDEO/AUDIO DATA NETWORK SERVICE PROCESSOR ENDSYSTEM A objb obja RIO APIC HIGH - BANDWIDTH BULK DATA ENDSYSTEM B Open ATM Signaling & Control SIGNALING SERVER WUGS OBJECT LOW LATENCY CONTROL HIGH-PERFORMANCE NETWORK ELEMENT USER KERNEL WEIGHTED FAIR QUEUING NSP PORT CONTROLLER ACTIVE NETWORK RESOURCE CONTROLLER SWITCH CONTROLLER SWITCH MANAGER SC WUGS ATM PORT INTERCONNECT CONTROLLER (APIC) NATIVE ATM BYPASSES THROUGH APIC RESOURE MANAGER CLASSIFIER/ ROUTER HIGH-PERFORMANCE NETWORK ELEMENT QOS MANAGER SMART PORT CARD IP FLOW CLASSIFICATION AND ROUTING SWITCH STATUS CHANGE NETWORK OPERATIONS CENTER - ENDSYSTEM ENDSYSTEM C MANAGE RIO APIC MIB MULTICAST HIGH -BANDWIDTH VIDEO/AUDIO GET STATS RECONFIG 24 Washington Un Audio/Video ing Flow data End-point Interface (Source) Control Object Interface Control Object Flow data End-point (Sink) Control And Management Objects Adaptor POA POA Adaptor ORB CORE MULTIMEDIA STREAM Flexibility Efficiency Uses CORBA for control messages and properties Sockets for data transfer to get high performance Concluding Remarks Researchers and developers of distributed, real-time applications confront many common challenges e.g., service initialization and distribution, error handling, flow control, scheduling, event demultiplexing, concurrency control, persistence, fault tolerance Successful researchers and developers apply patterns, frameworks, and components to resolve these challenges Careful application of patterns can yield efficient, predictable, scalable, and flexible middleware i.e., middleware performance is largely an implementation detail Next-generation ORBs will be highly QoS-enabled, though many research challenges remain 26