Batch analytics - Tracking UBE performance in JD Edwards

Tracking and comparing batch performance.

This is an update on my previous post https://shannonscncjdeblog.blogspot.com/2019/07/ube-performance-suite-with-dash-of.html on historical batch performance analysis.

The batch window can sometimes be an enigma wrapped in a riddle.  Mapping the F986110 to F986114 and lots of complex queries (like in my blog!!) to work out what is going on.

It’s hard when you might make changes to a certain UBE or to some central technology (batch server, database server) and you want to justify your existence.  You want to tell your managers how good you are and how many CPU cycles you’ve saved by implementing said changes?  Yeah, we all want that.

There are a couple of challenges with this approach, firstly that you delete your WSJ history and therefore probably do not see the full picture.  Perhaps a user even deletes their own history (I heard that this occurred once, a few years ago.  That user is now a consultant!  But also, really hard to summarise all of the jobs, with rows processed etc etc, and try to get some real results.

Then my team came up with ERP analytics for batch, where you can see all of this information -  no matter what you are doing with the history.

You can see that I’m using a fictional JDE company, but not a fictional company.  I try to keep my blog posts relevant on a number of levels.  www.watsacowie.com.au – get there!

Back to my technical blog.

This is fairly boring report.  You can see all of the UBE’s run for ever, all servers and all environments.  You can refine this with easy to use drop downs, that allow you to select a certain server or date range. 

Pretty cool, yeah, but look when you put a date range:

Then the magic happens

You can see that the report has automatically calculated the previous period dates (in this case another week further back) and has run a side by side compare for all of the UBE's.  This has compared about 180,000 records in seconds and gives me a point in time compare of this week over last week.

We can no see if any reports were processing WAY more data, or taking way longer or running more times that the previous week.  This is good information for understanding a change in performance and then using some of our other reports to work out where that performance problem is.

All these reports can be simply hosted on an E1page in JD Edwards, to that you can see your batch history all of the time and understand why things are performing like they are.

If you are interested in having this configured for your JD Edwards, then please reach out on linkedin or email me (you can work out my address, my first name "." my surname at fusion5.com.au - I'm trying to deter the bots)  we have a number of clients running with 1,000,000’s of historical batch records.  We automate the delivery of historical period comparisons so that they can know of any issues and act upon them.

Of course we are looking to augment this offering with two significant additions:

1. AI over the top of job status, rows processed and execution time.  This is going to perform anomaly detection over those dimensions and email you when something has gone wrong.  This has a far deeper understanding of the success or failure of the job, as it’s not just D or E.
2. Storing the actual PDF files also.  If you are ever worried about purging your WSJ – don’t.  We can carve off the jobs and the data and have this available to you in the EXACT same format as you see above.  A simple click of a link will allow you (if you have permission) to download the PDF.  This will keep your WSJ and PrintQueue directories (PDF database files) in check much better.  You also don’t need to worry about the auditors asking for reports.
3. Store the logs and CSVs
4. Make a portal for your customers to get their files!