The types
of jobs and services that you can store and run on a DataFlux Data
Management Server include real-time data services, real-time process
services, batch jobs, and profile jobs.
Real-Time Services Data Services
Real-time data services
are designed to quickly respond to a request from a client application.
Real-time data services can process a small amount of data input from
a client, or it can retrieve and deliver a small amount of data from
a database. Real-time data services are executed by the DFWSFC process,
as defined in
Manage the DFWSVC Process.
Real-Time Process Services
Real-time process services
accept input parameters only from clients, to trigger events or change
a display. Real-time process services are executed the DFWFPROC process,
which runs a WorkFlow Engine (WFE),
as defined in
Manage the DFWFPROC Process.
About Real-Time Data Services and Real-Time Process Services
If a real-time data
service or real-time process service fails to terminate normally,
then the service is terminated when the client connection times-out.
Real-time services
are stored in install-path\
var\data_services | process_services.
Batch Jobs
Batch jobs are designed
to be run at specified times to collect data and generate reports.
Batch jobs are not intended to provide real-time responses to client
requests.
Batch jobs are stored
in install-path\
var\batch_jobs.
Batch jobs, like real-time
process services, are run by the DFWFPROC process. You can pass input
parameters into batch jobs, but not any actual data.
Profile Jobs
Profile jobs are designed
to analyze the quality of specified data sets. Profile jobs are handled
as repository objects. They are required to reside in the Data Management
Repository. When you run a profile job, the server finds the job in
the repository and then starts a new instance of the DFWFPROC process. The
requested profile is then run by ProfileExec.djf, which resides in
the same directory as the repository. For more information about the
Data Management Repository,
see
About the Repository.Unlike batch jobs,
you cannot grant unique user permissions for profile jobs since they
do not have associated object-level access control. To learn more
about permissions,
see
Manage Permissions.When you install a
new version of the DataFlux Data Management Server, you are required
to import all of your profile jobs into a new Data Management Repository.
For more information about importing profile jobs,
see
Configure and Migrate after Software Upgrades.
Note that in Windows,
it is possible to save a DataFlux Data Management Server job to a
directory, and then not be able to see that job in that directory.
To resolve this issue, save your jobs in a location that does not
use mapped drives. A Windows service is not able to access mapped
drives, even if the service is started under a user account that maps
those drives.
A
DataFlux Data Management Server cannot run a remote job or service
whose location is specified with a mapped drive letter, such as Z:\path\remote_job.ddf.
The server runs as a service under a SYSTEM account and no mapped
drives are available to such an account. Use a UNC path to specify
the location of a remote job or service, such as: \\ServerHostName.MyDomain.com\path\remote_job.ddf.
The following restrictions
apply to the name of a job that will be deployed to a DataFlux Data
Management Server. You should follow these restrictions for all jobs.
A job name can contain any alpha-numeric characters, white spaces,
and any characters from the following list:
The maximum length of
job names is 8,192 bytes. DataFlux Data Management Server will not
upload, list, or run a job name with characters other than those cited
above.
In UNIX or Linux, to
run a shell command in a job, use the execute() function, as shown
in the following examples. To run the command directly:
execute("/bin/chmod", "777", "file.txt")To
run a command through a shell:
execute("/bin/sh", "-c", "chmod 777 file.txt")The
preceding examples return the host authorizations for a text file.