Measuring OS Support for Real-time CORBA ORBs Christopher D. Gill - - PowerPoint PPT Presentation

measuring os support for real time corba orbs
SMART_READER_LITE
LIVE PREVIEW

Measuring OS Support for Real-time CORBA ORBs Christopher D. Gill - - PowerPoint PPT Presentation

Measuring OS Support for Real-time CORBA ORBs Christopher D. Gill Washington University, St. Louis cdgill@cs.wustl.edu 27 January 1999 http://www.cs.wustl.edu/ cdgill/words-99. f pdf, ps g .gz Sponsors: Boeing, CDI/GDIS, DARPA contract


slide-1
SLIDE 1

Measuring OS Support for Real-time CORBA ORBs

Christopher D. Gill Washington University, St. Louis cdgill@cs.wustl.edu 27 January 1999 http://www.cs.wustl.edu/

cdgill/words-99. fpdf, ps g.gz

Sponsors: Boeing, CDI/GDIS, DARPA contract 9701516, Lucent, Motorola, NSF grant NCR-9628218, Siemens, and Sprint

slide-2
SLIDE 2

Christopher D. Gill Real-Time ORBs

Advantages of CORBA

  • www.cs.wustl.edu/ schmidt/

ORB-endsystem.ps.gz

Goals of CORBA

– Simplify distribution by automating

Object location

and activation

Parameter

marshaling

Demultiplexing Error handling

– Provide foundation for higher-level services

Washington University, St. Louis 1

slide-3
SLIDE 3

Christopher D. Gill Real-Time ORBs

Limitations of CORBA for Real-time Systems

  • Limitations

– Lack of QoS specifications – Lack of QoS enforcement – Lack of real-time programming features – Lack of performance

  • ptimizations

Washington University, St. Louis 2

slide-4
SLIDE 4

Christopher D. Gill Real-Time ORBs

But Can It Perform?

Low overhead

– context switch time – request-response latency – ORB/endsystem CPU overhead

Elimination of priority inversion

– context switch time – request-response latency

Predictability

– request-response jitter

Footprint

Washington University, St. Louis 3

slide-5
SLIDE 5

Christopher D. Gill Real-Time ORBs

Context Switch Time Measurements

Operating Context Switch Time,

sec

System mean (standard deviation) Suspend- Yield Test Synch Test Resume Test VxWorks 0.946 (0.041) N/A 1.62 (0.023) LynxOS N/A 5.42 (0.008) 5.96 (0.042) Windows NT 1.41 (0.036) 1.78 (0.021) 2.79 (0.110) Solaris 21.3 (0.569) 11.2 (0.900) 131.2 (0.613) Linux N/A 2.60 (0.023) 9.72 (0.187) Table 1: Context Switch Time Measurements

Washington University, St. Louis 4

slide-6
SLIDE 6

Christopher D. Gill Real-Time ORBs

Context Switch Time Measurements

Measured two context switches for two-way CORBA operation, as

expected

Contribution of context switch time, 1 to 10 sec, to latency is small Standard deviations are small, except on Solaris, where it may

impair performance predictability

Washington University, St. Louis 5

slide-7
SLIDE 7

Christopher D. Gill Real-Time ORBs

Priority Inversion Experiments

  • [P ]

C 0 C

  • 1

[P ]

C

Pentium II

2

1

C 3

3

[P ]

[P ]

[P ]

S

3

3

2

[P ]

S 2

2

I/O SUBSYSTEM

Server

1

RUNTIME SCHEDULER

[P ] S S

Object Adapter Servants

Client

[P] Requests Priority

[P ]

1

ORB Core

One high-priority client 1..n low-priority clients Server factory

implements thread-per-connection – Higher real-time priority for high-priority client – Lower real-time priority for all low-priority clients

Each connection links

client with its servant

Washington University, St. Louis 6

slide-8
SLIDE 8

Christopher D. Gill Real-Time ORBs

ORB Latency and Priority Inversion Results

200 400 600 800 1000 1200 Linux LynxOS NT Solaris VxWorks Operating System Two-way Request Latency, usec

20 Hz 10 Hz 5 Hz 1 Hz

Synopsis of results

– LynxOS and Linux provide lowest latency – LynxOS and VxWorks do not exhibit priority inversion; non-RTOS’s do – Therefore, deterministic QoS support requires a RTOS

Washington University, St. Louis 7

slide-9
SLIDE 9

Christopher D. Gill Real-Time ORBs

ORB Jitter Results

100 200 300 400 500 600 Linux LynxOS NT Solaris VxWorks Operating System Two-way Request Jitter, usec

20 Hz 10 Hz 5 Hz 1 Hz

Definition

– Standard deviation from average latency

Synopsis of results

– LynxOS and VxWorks show low jitter – Linux, Windows NT, and Solaris show unacceptable jitter

Washington University, St. Louis 8

slide-10
SLIDE 10

Christopher D. Gill Real-Time ORBs

Future Work

Latency and jitter measurements

– LynxOS offers low latency and jitter (high predictability) – VxWorks does also, with small number of low-priority clients; but does not scale well – Non-RTOS’ have high latency and/or jitter

ORB/OS operation throughput

– Fairly consistent results across platforms

ORB/OS CPU processing overhead

– Ranges from under 3 percent (Windows NT) to over 17 percent (VxWorks)

Washington University, St. Louis 9

slide-11
SLIDE 11

Christopher D. Gill Real-Time ORBs

Concluding Remarks

We can meet RT performance goals with CORBA, on RTOS’. Design and Implementation Challenges

– Dynamic and hybrid scheduling of requests and operations – Specifying QoS requirements – Distributed QoS and integration with real-time I/O Subsystems – Alleviating priority inversion and non-determinism – Reducing latency/jitter for demultiplexing – Reducing presentation layer overhead – Optimizing IDL compilers – Maintaining small footprint

Washington University, St. Louis 10

slide-12
SLIDE 12

Christopher D. Gill Real-Time ORBs

For Further Information

Performance Measurements:

– Demultiplexing latency:

www.cs.wustl.edu/

schmidt/GLOBECOM-97.ps.gz

– SII throughput:

www.cs.wustl.edu/schmidt/SIGCOMM-96.ps.gz

– DII throughput:

www.cs.wustl.edu/schmidt/GLOBECOM-96.ps.gz

– Latency, scalability:

www.cs.wustl.edu/schmidt/ICDCS-97.ps.gz

– IIOP:

www.cs.wustl.edu/schmidt/IIOP.ps.gz

Washington University, St. Louis 11

slide-13
SLIDE 13

Christopher D. Gill Real-Time ORBs

For Further Information

More detail on TAO:

www.cs.wustl.edu/

schmidt/RT-ORB.ps.gz TAO Event Channel:

www.cs.wustl.edu/

levine/research/JSAC98.ps.gz TAO static scheduling:

www.cs.wustl.edu/schmidt/TAO.ps.gz

TAO dynamic scheduling:

www.cs.wustl.edu/levine/research/scheduling/dynamic.pdf

ORB Endsystem Architecture:

www.cs.wustl.edu/schmidt/RT-middleware.ps.gz

Washington University, St. Louis 12

slide-14
SLIDE 14

Christopher D. Gill Real-Time ORBs

For Further Information

More detail on CORBA:

www.cs.wustl.edu/

schmidt/corba.html ADAPTIVE Communication Environment (ACE):

www.cs.wustl.edu/

schmidt/ACE.html The ACE ORB (TAO):

www.cs.wustl.edu/

schmidt/TAO.html These slides:

www.cs.wustl.edu/

cdgill/words-99. fpdf, ps g.gz

Washington University, St. Louis 13