MC
Hi,
I've been trying to play around PPM and discover it's capabilities. One of the capabilities that PPM boosts about is 'Automatic Process Discovery'. I've created few process tables in SQL server, and extracted the data from these tables in the form of event file using PPM's extract data program.
However, while importing the data into PPM's database, I had to model the process fragments using ARIS 9.6 and extract those fragments in a fragment file using PPM special report script.
My query here is.... is it necessary to model fragments and input the fragment file to PPM? Isn't there any other way where PPM can automatically generate fragments using event XML file? If no, then how is it automatic process discovery if one has to model the process fragments manually?
Please help me out understand the functioning of PPM as I'm new to this.
Thanks,
Mayank
Hello Mayank,
the general idea is, that you find traces of processes in your systems, databases, log files or wherever, that give you a hint, that a certain activity has taken place at a certain point in time.
As an EPC you could model this as an "atomic" process "Event - Activity - Event", called a process fragment. Now what you have to do in practice is enabling PPM to discover which End Event of such an atomic process correlates to a start event of another atomic process. Usually you will find some identifier, maybe an order ID or whatever characterizes the process instance in the system the data originates from, and some timestamp(s). In places, where the events originate from different systems you may have to invest a bit more thinking to correlate the processes, if the IDs are not the same.
The fruit PPM boosts about is that it will correlate all the atomic process instances to complete process instances, which can be visualized, and analyzed. I. e. each process instance always takes only one path, however multiple instances may take different paths. So the visualization will show how many processes took which path and you can drill down to the exception instances, which took multiple loops, more time or whatever statistical peculiarities they may have shown. You will be able to derive process-level KPIs that you won't find in any single log-file or data source.
Regards, M. Zschuckelt
@M. Zschuckelt : ARIS PPM is a great tool and I can agree with most of your points, but from the project I´ve done so far, there are some restrictions which make customers not so satisfied with PPM in productive environment. Especially nowdays when other products don´t require so much implementation´s effort.
And when it comes to the discussed visualization of multiple paths there is an issue I faced recently. If you want to reveal which particular process instances belong to a path (edge) there is a no go. Customers could not get why PPM can show you a number of processes which form the edge, but it doesn´t allow you to drill down for details and reveal the bad performers (=the procesess instances) one by one, how silly. You can do so only if you are interested in these numbers/probabilities on a function or an event. There you can easily do a drill down without any problem. In this particular case it does make a sense to see the bads under the edge, so please do not start a discussion if it is the best approach or not :)
@ Maya Chan : there is an option to reduce time consuming process of defining the fragment, because you can "model" your process fragments by editing XML file directly, so you don´t need to open Aris Architect at all. Or maybe you can simplify it with "Catch all" fragment´s definition. You should find the option in CTK.. But you are right the automatic process discovery is not so automatic :)
