Due to a vast emergence of distributed software based on CORBA, a clear need arises for tools to ease the development, debugging and profiling of such complex distributed software. Existing tools were not satisfactory for our intentions because they show only a sequential packet based view of the CORBA communication, they need instrumentation of the software under test or they only provide a single sniffing point. As a result, we developed our own tool for CORBA monitoring which supports at the user-end a GUI which gives a clear overview of the present objects and the communication between these objects and which is at the low level end supported by distributed sniffing of network traffic at different points in the network. Furthermore, sniffing is done by (high end) Linux PCs with the MIT Click Modular Router package which makes it possible to let the network cards work in a polling mode instead of in interrupt mode to avoid the receive livelock problem. As a result, fully loaded gigabit Ethernet links with a minority of CORBA traffic (which is, for example, the case if CORBA is used in network management) can be sniffed without dropping any GIOP packet. Future work includes further investigation of the profiling part by letting the monitor automatically create statistics about the recorded communication and in a further step this information can then be used for automatic object distributing and load balancing of CORBA objects based on monitoring input.
[1]
Massachusett Framingham,et al.
The Common Object Request Broker: Architecture and Specification Version 3
,
2003
.
[2]
EDDIE KOHLER,et al.
The click modular router
,
2000,
TOCS.
[3]
Bart Dhoedt,et al.
A generic end-to-end distributed QoS management architecture and its application to IP-DiffServ over a WDM access feeder network
,
2002,
NOMS 2002. IEEE/IFIP Network Operations and Management Symposium. ' Management Solutions for the New Communications World'(Cat. No.02CH37327).
[4]
K. K. Ramakrishnan,et al.
Eliminating receive livelock in an interrupt-driven kernel
,
1996,
TOCS.
[5]
Filip De Turck,et al.
Design of a generic platform for efficient and scalable cluster computing based on middleware technology
,
2001,
Proceedings First IEEE/ACM International Symposium on Cluster Computing and the Grid.