How To Monitor Multi-System Operational Processes

Organizations use processes to manage their business. How each process gets implemented falls into four main categories.

  1. Paper-based
  2. Email + Spreadsheet
  3. Multiple Systems
  4. BPM Software

Organizations have invested a lot of time and money in processes that fall into category 3. Bringing all those processes into a BPMS is not an easy or quick job. I have seen a single process that goes through 20+ systems from start to end.

Let’s take an example of order fulfillment, a fairly common process. The diagram below shows the lifecycle of an order as it moves through various systems. Most of the times you will find the core business of an organization distributed into multiple systems in a similar manner. Work, of course, gets done, but at any given time you cannot find exactly what is the current status of the process. And to generate even simple reports, data first needs to be consolidated from all the systems manually or through some sort of batch job.

 

Adeel Javed - How To Monitor Multi-System Operational Processes

So, how do you monitor a process that resides in multiple systems and ensure that it is running optimally considering it does not provide a single view? You build a shadow process.

The idea of a shadow process is simple; you design the end-to-end process with all major milestones without actually implementing all the functionality in a BPMS.

Implementation Steps

Here are high-level steps for implementing a monitoring system using shadow processes.

  • Define process with major milestones
    • Level of milestone granularity is directly proportional to effort required
    • Use milestones that make sense for reporting

Adeel Javed - How To Monitor Multi-System Operational Processes

  • Find a common id that will allow you to uniquely identify an instance of the process in all systems
  • Build services that either receives events from external systems or fetch data from external systems
  • Build services that advance the process instance based on event data

Adeel Javed - How To Monitor Multi-System Operational Processes

  • Log key event data e.g. instance id, business data, start time and end time etc. These can then be used by any reporting tool to generate meaningful insights

Adeel Javed - How To Monitor Multi-System Operational Processes

Compared to implementing all the processes in a BPMS, this type of monitoring solutions is much quicker to implement, provide end to end process view and helpful insights for future optimization.

 

This article was originally published on AdeelJaved.com.

 

pixelstats trackingpixel

By Adeel Javed @ AdeelJaved.com | June 15, 2017

2 Responses to How To Monitor Multi-System Operational Processes

  1. Can you elaborate on these points?

    Find a common id that will allow you to uniquely identify an instance of the process in all systems
    Build services that either receives events from external systems or fetch data from external systems
    Build services that advance the process instance based on event data

    You say it’s easier/quicker, these seem like significant activities that require quite a bit of effort.

    Thanks, Rich

  2. Hi Rich,

    In most processes whenever and however, a request flows from one system to another there is always a unique id that can be used to correlate e.g. order id, it will stay the same in most systems, all systems might add their own ids, but there will always be rules to match the same record in two different systems.

    You are right these are significant activities, but as I mentioned in the blog quicker/simpler to implement compared to implementing all the processes in a BPMS. If you were to take these processes that currently run in multiple systems, and implement them in a single BPMS, that would be a huge effort.

    Hope that clarifies.


    Adeel

Leave a Reply

Your email address will not be published. Required fields are marked *