View previous topic :: View next topic
|
Author |
Message |
aniceg Warnings : 1 New User
Joined: 04 Apr 2008 Posts: 10 Location: chennai
|
|
|
|
Is it possible to get OPC graph in a printer or in local machine from mainframe |
|
Back to top |
|
|
Lowell Fultz
New User
Joined: 14 Oct 2008 Posts: 1 Location: US
|
|
|
|
Isn't OPCGraph/TWSGraph a 2 piece product ? One for the host, one for the PC ?
There should be a GUI interface with which you view flows and you can print from that GUI. |
|
Back to top |
|
|
saturn
New User
Joined: 30 Jun 2006 Posts: 9 Location: Germany, Eschborn
|
|
|
|
I know only the graphical output from the z/OS TWS software (OPC). I think it is not useful to print that on a printer. The quality is pretty bad. The problem is that the software can only display schedules with few jobs in a useful quality. If you have complex work-flows you need a magnifying glass to see and find your jobs.
I think TWS Graph is a third party product which can do this work in a better way. But it is a commercial product and you have to spend money.
If you have time and fun you can do that work by your self. Have a look at JOb Schedule report from OPC. Currently I am working on a web base application. It is relatively easy to create flow chart with many jobs with GraphViz.
The only problem is that you need to process the scheduling data from IBM's TWS workload scheduler. I use a report printout from the long term planing which is provided by IBM data center. A REXX parser creates the needed data source from the text based report. Much better would be to use the original VSAM datasets from the scheduler.
If somebody has experience or knowledge about the VSAM layout I would be grateful for information. |
|
Back to top |
|
|
mtaylor
Active User
Joined: 20 Feb 2009 Posts: 108 Location: Kansas City
|
|
|
|
Saturn: I worked on a similar project using Graphviz a few years ago and even it's layout can't handle schedules with hundreds of jobs. When the preds/successors are poorly defined without knowledge of the over all schedule structure the graph quickly becomes spaghetti. Also for such graphs, Graphviz generates static output so navigation can be difficult or impossible unless one implements there own viewer using Graphviz data.
I've spent a lot of time on this so pm me if you want to discuss it further. |
|
Back to top |
|
|
Anuj Dhawan
Superior Member
Joined: 22 Apr 2006 Posts: 6250 Location: Mumbai, India
|
|
|
|
mtaylor wrote: |
I've spent a lot of time on this so pm me if you want to discuss it further. |
Unless it's get too site specific -- request to discuss that here, hopefully that would help many of us.
Have a good one,
Regards, |
|
Back to top |
|
|
saturn
New User
Joined: 30 Jun 2006 Posts: 9 Location: Germany, Eschborn
|
|
|
|
Hi mtaylor, you are right. Graphviz is only one part of my solution. I use Graphviz to create for each schedule a single page which contains the operations (jobs) as a single flowchart. The output is clearly arranged (not spaghetti like). I also add the predecessor and successor schedules into the graph. I use postscript as Graphviz output format. In the next step I add so called PDFMARK tags to enable hyperlink navigation between the pages (dependent schedules). I do this part with a perl script. I agree this can't be done by the Graphviz software.
At the end I merge all postscript pages to one big file and include the toc information also as PDFMARK code. Than it is easy to convert that ps file to pdf. Our smallest image contain about 3000 schedules. For that I have a pdf file with 3000 pages. The linkage works fine in PDF reader. The print quality is very good because pdf is vector based. The biggest images contain 18000 schedule. The resulting pdf is very big and not easy to handle. It is better to split the files in parts each ca. 3-5000 pages.
There is one limit: The bounding-box for the pdf format is limited. From our 80000 schedule (all german mainframes) are only about 10 affected by that limit. For example time nodes (e.g. beginning of batch at 20pm). But this not really a problem.
In my new project I use bitmaps with map files for navigation. The map files can be created by Graphviz without further manipulation.
I do the programming mostly in my free time because coding is not part of my current job. I am working in the mainframe production management of a big bank in Germany. The results are useful for several departments of my company. I also use the graphical workflow report very often in my work, because this is the documentation of the whole mainframe scheduling. I can get the same information from the OPC, but my visual report easy to understand and quicker to use. |
|
Back to top |
|
|
mtaylor
Active User
Joined: 20 Feb 2009 Posts: 108 Location: Kansas City
|
|
|
|
That sounds very interesting. What is the average number of jobs per schedule? I used Graphviz to generate diagrams for two schedules, one with about 250 jobs and the other with about 1100 jobs. When viewed as a pdf and sized for printing as a single page both were just smears of black. Plus using Windows tools, zooming and scrolling would lock up the machine for several seconds.
So I suspect you have many schedules with (relatively) few jobs. Is that correct? |
|
Back to top |
|
|
saturn
New User
Joined: 30 Jun 2006 Posts: 9 Location: Germany, Eschborn
|
|
|
|
The average is only 6 jobs per schedules. I think we are not running such big schedules. More than 100 jobs in one schedule (=application in opc scheduler) is a lot. Makes that sense? We also run large batch processes, but they consists of different dependent schedules.
Than you have more a problem with the pager size than with the results of the Graphviz output. |
|
Back to top |
|
|
mtaylor
Active User
Joined: 20 Feb 2009 Posts: 108 Location: Kansas City
|
|
|
|
The scheduler I was using was ESP. In OPC, do dependencies exist between applications/schedules? Then you could diagram two levels: jobs inside an application, and applications that depend on each other.... |
|
Back to top |
|
|
saturn
New User
Joined: 30 Jun 2006 Posts: 9 Location: Germany, Eschborn
|
|
|
|
Yes in OPC/TWS you have so-called applications (means job workflows or job schedules). Jobs inside the application can depend on other jobs from predecessor applications. In my visual report the predecessor and successor schedules are located on different pages. But this is not a problem because the dependent schedules are connected via hyper links. This is like a "walk-through" the batch processing.
How works ESP? |
|
Back to top |
|
|
mtaylor
Active User
Joined: 20 Feb 2009 Posts: 108 Location: Kansas City
|
|
|
|
ESP has the concept of 'application' which is a schedule of jobs. I believe the applications can be sequenced also. Iirc, the jobs in different applications could be preds/successors to jobs in other applications so it was very messy. Also the batch system I was working with had only two applications; they just put all jobs into these two. So a walk through was basically impossible. |
|
Back to top |
|
|
|