Introduction
In a more and more complex world of applications running on
different Open System servers throughout the network, system managers
and end users do need a powerful tool to manage the workload of their
environment. Even if applications tend nowadays to do their work more
and more interactively, payroll systems, database maintenance,
inventory control, ... need background jobs.
The standard Unix facilities "AT" and "CRON" can do this.
Unfortunately these lack some functionality and a lot of user
friendliness, e.g. good planning and scheduling, careful execution,
thorough monitoring and analysis. A job scheduling system need to match
these criteria. In order to control the complexity of scheduling,
initiating and monitoring the workload in your environment, Ockham
Technology has developed JOBWARE, the powerful tool for workload
management.
JOBWARE empowers the System Manager to overview (networked) systems
with all its queues, scheduled and completed jobs. Full journalling,
recovery, reporting and alarming facilities assist the user to keep the
system reliable and complete the job whenever and wherever needed.
Moreover JOBWARE allows you to set up the control of your distributed,
heterogeneous servers from a single point or to distribute the control
itself all across the network. And, of course, client/server becoming
the reality, JOBWARE has been designed appropriately, offering all the
practical benefits this approach can deliver.
The next paragraphs explain the benefits of JOBWARE. Discover how easy
scheduling can be.
"DAEMON" : The heart of the matter
The heart of the matter, the motor of the JOBWARE product, is
the JOBWARE “Daemon”. This background process decides what has to be
done, when and depending upon which conditions. This know how is
derived from the definitions in the internal database and user
initiated actions.
The Daemon pre-schedules these activities every day. This way a check,
which jobs need to be executed that day, will be performed according to
the specifications of the job.
For security reasons, the Daemon can only be inspected and controlled
by the Manager (see following).
Users
JOBWARE users can be of different types, according to the
level of capabilities they have been allowed. Capabilities regulate the
security of the system. Only the Manager controls the JOBWARE system
with its queues, its (other) users and the Daemon. Operators control
the job processing. Users are allowed to schedule and control only
their own jobs.
This hierarchical structure of capabilities enables the System Manager
to set up a strongly secured system where "vital", pre-scheduled jobs
are being protected from end user intervention. Besides it helps the
operator to intervene only where and when necessary.
A User typically wants to launch his jobs and to have an overview of
the status of it. Moreover he needs an overview of the scheduled jobs,
those running, the priorities, ... Maybe some jobs need to be
rescheduled or released or another configuration is needed. And how
about the predefined jobs? How do they interfere with the user launched
jobs?
After jobs have been launched, users can have at all times a general
overview of all these jobs and execute several actions on these jobs.
Types of actions that can be performed are:
-
modifying the job status : delete, abort,
defer, release, suspend or resume a job
-
transferring jobs between queues, even
between different machines by means of the virtual queues
-
modifying the priority of the job in
the queue
-
modifying the schedule date
-
modifying the prerequisites of a job
All these actions give users of JOBWARE a large
flexibility in managing their launched jobs and helping controlling the
CPU allocation. Besides they are implemented in such a way any user can
handle them in the interface he likes the most.
User interfaces
The way Users, Operators and the Manager handle the
JOBWARE product is defined by different user interfaces, thus leaving a
wealth of choice to them. The users decide independently from their
colleagues what user interface they like to work with. The product is
available in a command line interface, with character based windowing,
under MS-Windows. This means that a JOBWARE Operator might prefer the
command line interface, while the Users themselves within the company
prefer Windows '95.
Jobs
JOBS are either predefined or initiated by a user. User
initiated jobs are "ad hoc" jobs, jobs a user fires off on the spot.
Predefined jobs are waiting to be scheduled at a regular basis, maybe
depending upon certain prerequisites. These conditions (date, time,
calendar, queue, prerequisites, ...) control or trigger the repetitive
character of the job. Special scheduling functions can be added to
control the execution flow and/or order of the job. In this way the
user may sequence, exclude or synchronize jobs and the job definitions
can be analyzed at all times.
JOBWARE has extensive calendar functions. Free naming of
groups of days (e.g. to specify company holiday periods, closing dates
for bookkeeping, ...) help to plan predefined jobs in a much easier way.
Here is an overview of the extensive launching options of JOBWARE.
Options for user initiated jobs:
-
to execute: a filename or the
command(s)
-
date/time: they are only limited
to a specific day and time
-
queue
-
priority
-
name: used as aliasname for a job
in the different views and reports
-
state options: jobs can be put in
a “defer” or “hold” status for later release
-
mail and alarming options: if the
job has finished, mail will be sent to a user, eventually acting as an
alarm trigger in case of problems
Options for predefined jobs:
-
to execute: a filename or the
command(s)
-
date/time: can be a specific day
and time or based on a calendar or a complex date and time definition
-
queue
-
priority
-
name: used as aliasname for a job
in the different views and reports
-
state options: jobs can be put in
a "defer" or "hold" status for later release
-
mail and alarming options: if the
job has finished, mail will be sent to a user, eventually acting as an
alarm trigger in case of problems
Extra options for predefined jobs are:
-
username
-
groupname
-
prerequisites: these are filters
that control the execution of the job. Possible filters are:
-
existence of
-
size of files
-
preventing jobs from running
simultaneously
-
chaining jobs
Reports
A scheduling system needs extensive reporting to trace
events happening and happened thus enabling appropriate corrective
actions to be undertaken. JOBWARE has three different extensive
reporting facilities:
-
journals for history logging. They
summarize all information on finished jobs. The reports of each
executed job show extensive and detailed information concerning
jobname, jobid, exit status, etc.
-
job analysis reports for
simulations. They assist in creating scenarios of the use of predefined
jobs and analyzing this report assists in improving the efficient use
of system resources.
-
reports on the Daemon activity for
recovery and system management. They indicate when the Daemon has
started and when the preschedule has been performed.
Recovery
Full recovery facilities assist the System Manager in
ensuring a reliable system. The Daemon indicates when it is not running
on a system. It can be restarted in a two ways:
-
recover mode: this mode recovers
all jobs which have been launched but failed to execute due to a Daemon
stop (system crash, disk fault, ...). In case a job has been aborted
during system shutdown, the Daemon adds it automatically to the
appropriate queue. Besides a preschedule is being performed for every
day the Daemon has been dead.
-
initialize mode: this mode
performs only a preschedule, without a recovery of the aborted and not
executed jobs.
Networks
In a world of distributed applications, a networked
scheduling system is a must; the scheduler needs to control, manage and
alarm different server nodes. JOBWARE has been designed to perform this
task appropriately. Its easy to configure network database contains all
the hosts where JOBWARE is installed. This database enables to specify
virtual queues, logical queues which are defined on a host, but which
are physical on another host. In this way users define virtual jobs,
jobs to be executed on other nodes within the network.
Besides JOBWARE has been developed according to the client/server model
and consists of two separate and interacting processes: the client and
the server process.
The client is the process that launches a job, either user initiated
from the user interface or being specified by a predefined job. The
server is the Daemon that sets up a process to execute the job and that
returns to the client the necessary information.
Because of this client/server structure, multiple JOBWARE nodes in a
network offer the user transparent access to all scheduling
functionality on every host in the network.
Application integration
To make life easier for the developers, JOBWARE provides
specific commands and Client/Server API’s.
JOBWARE commands can be used to launch jobs from any UNIX application.
The JOBWARE Client/Server API’s can be used in PC applications to
launch jobs on the UNIX server, assisting this way in creating real
Client/Server applications OR they can be used to create a specific
user interface.
This way, JOBWARE is an open solution for all job management issues in
applications across a network.
Available platforms
Following platforms are supported:
-
HP-UX
-
IBM AIX
-
DIGITAL Tru64
-
SCO-UNIX
-
SOLARIS
-
WINDOWS server systems.
References
Multiple companies all over the world do benefit from
the JOBWARE product. Some major references are: L’Oréal,
Bekaert, Ericsson, Prestige & Collections, Logica, Finnish Telecom,
Preferred Assurance, ...
Benefits
JOBWARE provides numerous productivity benefits both for the operation
staff and for the end-users:
- for the operations staff because:
-
they can quickly and easily manage
a massive number of jobs simultaneously
-
JOBWARE handles the workload, so
they can concentrate on exception handling
-
they can assure an optimized
turnaround time and throughput, because JOBWARE automatically controls
and processes the background production workload
-
from one single point, they have
an overview of all jobs on all the systems in the network, not matter
how complex
-
they can assure an improvement of
the system reliability
- the numerous reports simplify the auditing>
-
for the end-users:
-
their sessions are not any more
blocked by running processes interactively; these can now easily be
scheduled in background
-
they have a free choice of
interface, thus minimizing learning efforts
-
they can consult at all times a
complete overview of what’s happening.
|
|