Workflow Management System in BeesyCluster 183
Note that the target Java EE server may have access to the location of
input data to a particular workflow application as the database with access
data can be shared among Java EE instances (Figure 6.20). All Java EE
servers can use the same database system, which may be composed of several
servers. This method ensures high, both intra- and interserver, scalability of
workflow application execution.
The current implementation of the workflow management in BeesyCluster
supports handling of failures of services [66] through successive launching of
the service with an imposed timeout. If not succeeded, the service is assumed
to be inaccessible or unable to return results and rescheduling is triggered
using the ...