Evaluation - Autonomics Development: A Domain-Specific Aspect Language Approach - page 110

RemoteJ generated client
Multicast client
Network
Multicast server
RemoteJ generated server
Figure 6.16. Multicast recovery. A multicast server is embedded
in the RemoteJ generated server using the serverPlugin state-
ment, which allows the embedded code to run in a separate thread
in the server application. A user-defined recovery routine is then
able to interact with the server thread.
For our evaluation we implemented a simple system where a client in need
of an alternative server broadcasts a request on a multicast network 1 . The first
responding server, providing it is not the current server, is chosen.
For our implementation we embedded a simple multicast server in the Re-
moteJ generated server using RemoteJ's serverPlugin capability, described in
Section 5.3. Upon application startup, the MulticastServer server plugin is
started in a separate thread and waits for a broadcast on a well-known multicast
group and associated port number whose configuration may be changed by altering
the following property items for both the client and the server:
• remotej.multicast.sendPort. Theportusedtosendtherequeston 2 .
• remotej.multicast.receivePort. Theportusedtowaitforaresponse.
1 A multicast message is a message that is sent to a group of hosts that subscribe to an address
group, thereby allowing broadcasts to be sent to that group only rather than to an entire network.
Address groups are defined in the range 224.0.0.0 through 239.255.255.255 for the IP protocol
[96].
2 Note that the client send port must equal the server receive port and the client receive port
must equal the server send port.