For Managing Workflows Changes, “It’s Only Weird If It Doesn’t Work” Doesn’t Work So Well

With all the technology available today, there is at least one area where weird and unusual processes still rule. For some reason, even though batch workload like payroll, inventory and supply chain management account for about 70% of all business processing done on commercial computers, new methods like DevOps are completely ignoring this critical function.

So how do organizations manage changes to their batch workload? Well, we asked a bunch of them and the answers range from the sublime to the ridiculous. Some use the tried and true “pretty please” approach. A developer or someone else who needs some jobs built, calls his or her friendly scheduler and says something like ”please do me a favor” and add this job or change that. Some have built incredibly elaborate systems of forms that have to be filled out and submitted. Until every last bit or information required by the form is acquired, schedulers cannot do anything. Sometimes, the form is filled our correctly on the first try (not very common) and sometimes the request goes back and forth and back and forth until people start getting blue in the face and making some very impolite gestures. Many companies fall somewhere in between and use some kind of request mechanism based on Excel spreadsheets, Word documents or email. One company we spoke with demands a Visio diagram of the desired results and then schedulers manually cut and paste from the Visio into the forms defining the workflow. One company told us they don’t really have a process. When somebody wants a job, they have to hunt down the people who can provide that service and each request is a negotiation that makes the Middle East peace process look trivial.

Of course, in all cases, the changes that have to get made DO eventually get made and thus the reference to “it’s LuckySocks.jpgonly weird if it doesn’t work”.  However, the fact that people have to jump through hoops and do unnatural things IS weird and our definition of “work” needs to be seriously revised. First, the entire process needs to be brought out into the light of day and recognized for what it is; inefficient, problem-ridden, expensive and frustrating. Second, the lack of automation introduces errors that cause production failures because the actual job definitions that make it into production are frequently built manually from scratch with no standards validation. Third and perhaps most important, the applications that businesses depend on to keep them competitive or to gain an advantage or address customer needs are being delayed.

You may think what’s the big deal? Let everyone use the scheduling tools you already have and problem solved. Well, if it was that easy, everyone would have done that a long time ago. The problems with that approach are that you can’t tell all your application developers or business analysts to become workload automation experts, the IT Operations and Scheduling folks would freak out (rightfully so) at the thought of hordes of users having free and unfettered access to your production environment and your auditors would probably read you the riot act. Even if you could pull of this minor miracle and get such an action approved, your auditors wouldn’t be happy, application developers wouldn’t be happy because they already have a day job and management wouldn’t be happy when production workload failed due to errors introduced by novice or casual users.

The solution requires some new capabilities that up until now have not existed. What’s needed is a solution that is simple to deploy and use by non-scheduling and/or casual users without extensive training. This must also be able to support local and site-specific standards and adoptable for different levels of users so that requests are as close to fully formed and ready to run as possible. It must be automated to eliminate the need for manual construction of workloads yet sufficiently collaborative to provide different levels of engagement between requesters and schedulers. Finally, it must be fully controlled and audited to ensure that the critical production environment is not compromised in any way.

With such a solution in place, developers and business analysts/owners will be able to define their requirements for application workflows quickly and efficiently. IT can implement these new business functions smoothly and confidently without impactimg the quality of their service delivery. The most efficient use is made of everyone’s time and most important of all, the business gains the benefits of this accelerated application delivery to provide new products and services that can ward off competitors or expand market share.

So next time you have some workflows to create, you can pull out your lucky socks and hope for the best or come back here in a few weeks to learn about a better way.

These postings are my own and do not necessarily represent BMC's position, strategies, or opinion.

Share This Post


Joe Goldberg

Joe Goldberg

Joe is a solutions marketing consultant at BMC with over 30 years of technical marketing and management experience. He specializes in large systems architecture, systems management software, and enterprise solutions and has a strong interest in emerging markets.