DocumentCode
2558561
Title
Evaluating overhead and predictability of a real-time CORBA system
Author
Zou, Jianfan ; Levy, David ; Liu, Anna
Author_Institution
Sch. of Electr. & Inf. Eng., Sydney Univ., NSW, Australia
fYear
2004
fDate
5-8 Jan. 2004
Abstract
Over the last few years, real-time CORBA are increasingly used in development of distributed realtime applications, in which real-time CORBA and operating system constitute the computing and communication infrastructure upon which the applications execute. However the amount of effort put into studying and evaluating the real-time characteristics of real-time CORBA is still inadequate. In this paper, we focus on examining the overhead and predictability of real-time CORBA, which are two of the most important concerns when building real-time applications. Firstly, we introduce the desired features in real-time applications and possible sources of unpredictability. Secondly, we evaluate a real-time CORBA implementation : TAO v1.2 on Linux kernel v2.4.18 over 100M Ethernet. It is observed that: 1) the overhead of TAO is acceptable, compared with the equivalent application using TCP/IP socket directly; 2) TAO presents rather predictable behaviors under light network traffic, but unpredictability of CORBA invocations increases significantly as the network becomes overloaded. Moreover, we identify that the major source of unpredictability of CORBA invocations is in the TCP/IP stack of Linux rather than in TAO itself.
Keywords
Linux; distributed object management; local area networks; real-time systems; transport protocols; Ethernet; Linux; TCP/IP; network traffic; real-time CORBA system overhead; real-time CORBA system predictability; Distributed computing; Ethernet networks; Kernel; Linux; Middleware; Network servers; Operating systems; Real time systems; TCPIP; Yarn;
fLanguage
English
Publisher
ieee
Conference_Titel
System Sciences, 2004. Proceedings of the 37th Annual Hawaii International Conference on
Print_ISBN
0-7695-2056-1
Type
conf
DOI
10.1109/HICSS.2004.1265651
Filename
1265651
Link To Document