Text preview for : B410 Introduction To JES3; Hood.pdf part of IBM B410 Introduction To JES3; Hood IBM share SHARE_61_Proceedings_Volume_1_Summer_1983 B410 Introduction To JES3; Hood.pdf
Back to : B410 Introduction To JES3 | Home
PAGE 2
iSHARE:
=
SHARE SESSION REPORT INTRODUCTION
61 B410 MVSN: Introduction to JES3 60
SHARE NO. SESSION NO. SESSION TITLE ATTENDANCE
MVS New Users David Wichmann HIO This presentation Is a brief overview of JES3. It Is assumed
that you have at least a general conceptual understanding of
PROJECT SESSION CHAIRMAN INST. CODE MVS, but no background In JES3.
HILTI, Inc., 4115 S. 100th E. Ave., Tulsa, OK 74011, 918-627-9711 X2857
It is not a comparison of JES2 and JES3.
SESSION CHAIRMAN'S COMPANY. ADDRESS. AND PHONE NUMBER
The main topics to be covered will be concepts. major
functions. and configurations. The flow of a standard job
through the system will be traced. Some of the important JES3
facilities will be examined. And finally. the user interface
will be surveyed.
The presentation is on the JES3 component of OS MVS/System
Product. otherwise known as JES3/SP~ The JES3 component of OS
MVS/System Product Is supported in the System/370 environment
as part of HVS/System Product Version 1 Release 3.1 and In the
System/370 Extended Architecture environment as part of
HVS/System Product Version 2.
INTRODUCTION TO JES3
First. let us examine the general functions that any Job Entry
Chester Hood Subsystem should possess. Then. we will see how JES3 relates
I'D
=
=
Maremont Corporation
Hail Stop 205
P. O. Box 1488
to HVS as it performs these functions.
The general job entry subsystem functions are:
Nashville. Tennessee 37202
(615) 360-1213 1. System Input Source. The job entry subsystem gathers jobs
and the i r related data from local and remote Input
devices.
2. System Output Fac 11 I ty. The job entry subsystem Is
lEVEL OF JES3: responsible for disseminating output to Its appropriate
MVS/System Product Version 1 Release 3.1 destination.
HVS/System Product Version 2
3. Job Awareness. The job entry subsystem provides control
Printed: 08-19-83 over the schedull ng and execut I on of jobs. JES3 provi des
total job awareness in terms of resource allocation for
all jobs in the system.
MVS users today can choose between JES2. which was derived from
HASP. and JES3. which was derived from ASP.
Under MVS. much of the job entry subsystem executes in Its own
address space along side of system address spaces such as the
Master Scheduler. started tasks. batch jobs. and TSO sessions.
The JES address space is normally non-swappabla and runs at a
high dispatching priority. Also shown Is the new optional JES3
Auxilliary Address Space.
3/E/LEJ/l
PAGE 3 PAGE 4
The JES3 confi gurat i on shown Illustrates .. any key elements. It JOB FLOW
also serves to point out some major JES3 concepts. A three
processor (triplex) installation is being depicted. Each
processor shares access to the sPool yolumes. One of the
processors is designated as the JES3 ~ processor. The A standard job under JES3 is scheduled through five segments of
other processors are known as JES3 .l..!!s!.l processors. process i ng and a 11 except a port i on of one segment wi 11 be
performed by a function that executes within the JES3 address
The global must be connected to each local by a space on the global processor. These functions are called
channel-tp-channel adapter (CTC or CTCA). Optionally, there Dynamic Support Programs (DSPs).
may be a CTC connection between the locals. The use of these
connections will be covered later. Any such configuration of
two or more processors connected by CTCs and/or shared DASD
fits the definition of a looselY coupled multiprocesslno INPUT SERVICE.
~
The global processor provides a sjnole system jmage of the Jobs are read from local and remote input devices attached to
entire complex of processors. All of a job's input stream data the global processor and placed on the spool. Input service
and all of its syste .. -printed output flow through the JES3 supports card readers. tape readers. disk readers. and remote
global. The global reads the job stream and writes it out to raaders in a remote job processing (RJP) or a network job entry
the spool for later schedul i ng and execut i on under MVS on (NJE) configuration. The first phase of input processing is
either the global or one of the locals. The data which is sent the reader. which is skipped by the internal reader. Jobs are
across the CTC consists of messages relating to the status. read from an input device and placed on the spool In batches.
ownership. and control of work on the queue. or the control of The main purpose of this phase is to separate jobs and their
the processor. data.
N The maximum JES3 configuration repr.esents a signl ficant amount The second phase is control statement processing. This phase
...
= of hardware resources. Performance and phys i cal
considerations may limit the actual maximum configuration for
an individual installation. JES3 can support the following:
builds JES3 control blocks. which will be used throughout the
life of the job. and modifies the defaults in the control
blocks with the information which was retrieved from the
control statements. These control blocks are then written to
1. Only one global processor (required) the spool.
2. Up to seven MVS local processors
3. Up to 31 shared DASD queue volumes
INTERPRETER SERVICE.
The function of interpreter service is to convert JCl
statements into scheduler control blocks and additional JES3
control blocks. The converter/interpreter (C/I) is used for
this purpose.
The converter reads the JCl from the spool and converts it into
internal text. If there are any cataloged procedures in the
job. they are resolved by using user defined proclib
concatenations.
Next the interpreter is used to convert internal text into
scheduler control blocks. Any catalog references by the job
are resolved at this time. From the scheduler control blocks
and the catalog information. a set of JES3 control blocks which
contain a complete profile of the job's resource requirements
is created. The scheduler control blocks. job setup table. and
job volume table are all written to spool.
PAGE 5 PAGE 6
If any JCL errors are detected (syntax and logic), only the Jobs are selected for resource allocation based on
output and the purge stages of the job wi 11 be scheduled. installation job priorities in first in/first out, or FIFO,
order.
Only those processors with sufficient device resources will be
MAIN SERVICE. considered for setup. MDS performs allocation on a
complex-wide basis for data sets, volumes, and devices.
Main service is a segment of job scheduling which collects This provides the ability to have complex-wide data set
three areas of JES3 processing which deal with the execution of integrity for data sets on shared devices via an enqueue-type
a job under MVS. The areas are: mechanism. This control is effected through a combination of
data set name and volume serial number. If two jobs request
1. Main Oevice Scheduling exclusive control of a data set, they would not be scheduled
- pre-execut i on setup for concurrent execution on the same (or different)
processors.
2. Generalized Main Scheduling
- controls workload However, if identically named data sets exist on different
volumes, MDS correctly recogni zes that they are separate,
Main Servicing unique resources and would allow concurrent execution of the
3.
- interprocessor communications two jobs. If the two jobs were scheduled to the same processor,
- job scheduling and termination then one would be delayed because the MVS data set enqueue is
still performed using only the data set name.
The abi lity to make a specific volume serial number unavai lable
MAIN DEVICE SCHEDULING for use for long periods of time is provided. MDS keeps track
of which volumes are currently available throughout the
~
c: The first processing a job will have done is Main Device complex.
N Scheduling CMOS). MOS controls the fetching, allocation, and
mounting of the direct access and tape volumes requested in the It knows which volumes are being used and how they are being
JCL of each job to be executed on a processor. MDS is broken UP used. In addition it will keep a volume mounted, which was
into the following areas: released from one job, if it is required by another job.
1. Volume Fetch Fetch messages will be sent to the appropriate library.
2. Device Allocation Mount messages can be sent to a console near the device to be
mounted.
3. Volume Mount processing
HDS then verifies that the volume which was mounted was indeed
Device Deallocation the requested one.
4.
There are several reasons for JES3 to do allocation: If all volume mounts have been satisfied, then the job is ready
to be scheduled for execution.
1. MVS allocation is on a step basis only (delay due to
resource unavailability) This early allocation of devices guarantees that all of a jobs
data set, volume, and device requirements can be met before the
2. MVS allocation does not know about other systems (requi red job is scheduled for execution.
volume in use on another processor)
Usually breakdown for a job occurs as job steps complete and
The required number of devices may never be available also at the end of the job. For dynamic allocations, breakdown
3.
wi 11 take place when the corresponding dynamic deallocation
Make the max i mum use of dev ices occurs.
4.
Have a job remain for the minimum amount of wall clock time Resources allocated to a job will be released when they are no
5.
once it has been passed to the system for execut i on longer required by the job.
PAGE 7 PAGE 8
MDS functions are optional and may be bypassed for a job as well conditions and the overflowing of prespecified output limits.
as on a complex-wide basis. Write-to-operator messages which a job issues are routed back
to the global processor to be logged in the jobs output and in
addition sent to console service.
GENERALIZED MAIN SCHEDULING.
Both common and dynamic allocations/deallocations are reviewed
After MOS. a job is placed onto the execution queue for the by JES3. If major decisions are necessary, then the matter is
scheduling of the job by Generalized Main Scheduling (GMS). dec i ded by MDS on the global processor.
GNS controls the workload and maximizes system throughput. in
addition to optimizing job selection for machine loading. GMS
provides a flexible framework for establishing priority
relations between job classes within groups and between OUTPUT SERVICE.
groups.
JES3 has provided the installation a great many controls over The output service function processes SYSOUT data sets. Output
job scheduling which will make it easier to manage the work can be destined for local or remote printers. punches. TSO. an
flow through the complex. Effective usage of these facilities external writer. or the internal reader.
can result in a reduction of the manpower which is required for
the tasks of scheduling, monitoring, and control over jobs. Output service consists of two phases: scheduling and writing.
Scheduling is done on the basis of output characteristics which
These facilities include: are known via JCL. JES3 control cards. or the SYSOUT class
table defined at initialization. These characteristics
1. Job priority include:
2. Interaction of jobs within a group 1. Data set pr i or i ty
N
c:::: 3. Processor dependency (explicit and implicit) 2. Data set dest i nat i on
~
4. Logical storage size 3. Device type
5. Class mix (limit one class by another class) 4. Forms name
6. Mix of differing I/O rates (low. medium. high) 5. Carr i age tape name
7. Initiator availability (by group. the main driving force 6. Print train name
of GMS)
7. SYSOUT class
8. Sequencing of related jobs by Dependent Job Control
8. 3800 char acter i st i cs
Spinoff. or immediate printing. of data sets is also supported
by JES3. Scheduling for these data sets occurs at the time they
MAIN SERVICING. are spun off.
The main servicing support function controls job execution A unique output service writer is started for each active
processing within the JES3 complex. Main servicing on the output device. At start time. the writer is associated with a
global processor interfaces with all other processors via a eTC given set of SYSOUT characteristics. The writer can be either
and with the global processor via SRB scheduling. Job hot or dynamic. Hot writers are started once by the operator
execution takes place on the global.or a local main processor. and wait for a specific type of work. Dynamic writers are
started and stopped under JES3 control as output to be
During execution on a local processor. the job reads SYSIN data processed appears.
sets from the spool and writes SYSOUT data to the spool using
JES3 data management routines. JES3 monitors some of the On tightly coupled multiprocessors. much of the writer
functions which a job uses. JES3 watches out for abnormal processing is dispatchable separately from the mainline JES3
PAGE 9 PAGE 10
processing, thus effectlively using both instruction stream FACILITIES
processors.
An external writer is a routine running in its own address
space which communicates with JES3 through the subsystem CONSOLE SERVICE.
interface and writes output to an output device. The TSO
output command uses this same interface.
Console service provides communication between JES3 and the
operators of the system. JES3 console support features
include:
PURGE.
1. Multiple functional consoles (virtually unlimited vs. 32
MCS consoles)
The final segment of scheduling through which all jobs must go
is Purge. This is a very important stage when all remaining 2. Coex i stence wi th MCS
resources associated with a job are released for use by other
jobs. The main resource held at this time is spool space. This 3. Subsystem interface
is also the point at which the final job accounting information
is wri tten to the SMF data set (record type 26). MCS commands (SVC 34)
WTO/WTOR (SVC 35)
WTl (SVC 36)
DOM (SVC 87)
4. Single system image for control of the entire complex
~
=
..... 5.
6.
JES3 and MCS commands from the same console
Extens i ve MCS and JES3 command language
7. User exits in console service
Shown is a JES3 configuration which has multiple functional
consoles. JES3 supports up to 96 unique destination classes
for messages as opposed to only 16 for MCS. This faci lity
allows an installation to locate its consoles close to the
logical work flow. Consoles may be located close to card
readers, printers, card punches, tape drives, disk drives, and
in the tape and disk libraries.
Messages pertinent to these operations can be routed to the
appropriate consoles. Each RJP work station has a logical
console for the control of its work flow. Functional consoles
are all attached to the global processor.
JES3 console support co-exists with MVS multiple console
support. JES3 consoles can be defined as JES3 consoles, MCS
consoles, of both. JES3 consoles permit the entering of JES3
commands to the global or to any of the local processors. MCS
support does not apply to remote consoles.
MCS consoles allow the entering of MVS commands to the
processor to wh i ch they are attached. JES3/MCS consoles allow
communication to both the global processor and the local
processor using JES3 and MVS commands.
PAGE 11 PAGE 12
Consoles can also communicate with each other. The need to NETWORKING
station an operator at each local processor can be reduced or
eliminated. This provides the desired single system image from
an operational standpoint. JES3 Networking is an important facility which allows
communication and interchange of job data between a JES3
All console message traffic in the JES3 complex is recorded in comp 1 e x and ot her remote J ES3 comp Ie xes, as we 11 ali J ES2 and VM
the system log (SYSLOG). In addition to the message text is a systems.
time stamp and in some cases the id of the associated console or
the system whi ch generated the message. Shown here are each of the types of systems wh i ch can
participate with JES3 as "nodes" in a network.
A JES3/MCS console with the proper authority can enter 28 MVS
commands and 18 JES3 commands. Four major levels of command Networking provides facilities to transmit:
authority define what level of control and monitoring is
authorized for a given console. Short forms are supported for Jobs
many commands. In the event of a console fai lure, the function SYSOUT Data
of the failing console can be switched to an alternate console. Messages
Many JES3 parameters which were defined at initialization time Commands
can be displayed via the INQUIRY command and changed by the
MODIFY command.
On graphi cs type consoles, commands can be assoc i ated wi th DEPENDENT JOB CONTROL.
program function keys (PFKs). These features illustrate the
human factors and RAS capab iii ties bu i 1 t into the support.
Dependent Job Control (DJC) is a general function In JES3 that
allows the installation to manage job sequencing. Job sequence
HI dependencies often occur because one job's output may be
c REMOTE JOB PROCESSING. another job's input. There may also be catalog dependencies.
CoIl These dependencies may be quite complex. An example is shown:
Another valuable JES3 facility is Remote Job Processing (RJP). 1. Jobs Band E are dependent on Job A.
Under RJP remote card readers, punches, printers, and console
function as logical extensions of similar local devices. RJP 2. Job Dis dependent on Jobs Band E.
supports both system network architecture (SNA) with
synchronous data link control (SDLC) line protocols and binary 3. Job Cis dependent on Job B.
synchronous communications (BSC) line protocols and terminals.
To control this kind of scheduling