Sequential Batch Processing via Data Process Control vs. Daisy Chain

The sequential batch processing capability has been improved in the Data Process Control. Historically, the Data Process Control has always had the ability to sequentially batch feeds together on the fly in the On-Demand panel by providing the end-user the ability to select multiple feeds. However, this ability was always limited, as the Schedule panel was incapable of doing this. Each feed had to be individually scheduled and staggered, and this led to potential collisions and/or process inefficiencies. This problem has now been solved by executing p_UI_SCHEDULE_AGTBLD_v1.3.sql script. This patch is a systemic part of the DBObjects 1.5.1 packaged with the 112 pack. Once this patch has been put in place, the Data Process Control will allow you to schedule a block of feeds together as a single daisy chained process. To do this, simply schedule the feeds to run at the exact same date/time, and the schedule panel will be intelligent enough to batch them together as one unified process, similar to the way the load panel works. See example screen shot:

Using the DPC to control the daisy chain is the MUCH more effective than the classic string together approach outlined here. The string together approach is not ideal, and should not be used except under certain specialized circumstances. The approach of hiding feeds in a single "umbrella" feed should never be done, for the following reasons:

1. It obfuscates process understanding from user. If a user understands the business process, and the feeds are clearly labeled to match the business process, there is no need to "hide" anything
2. Hiding feeds in a single umbrella feed creates confusion when something goes wrong. Inevitability in complex processes, something ALWAYS goes wrong.
3. Each process, if need be, should always be able to be run in isolation.
4. BPM Connect is unable to effectively run feeds within feeds. If you run a External Process in the data pipeline you can run as "IN" which will confuse the data process control, and mess the logging up. If you run as "OUT" you lose the ability to do sequential dependency.

The Daisy Chain approach is still available is is not going away, however, it should be used in exception scenario and not as a standard. For the DPC sequential batch approach to be effective, two things need to be in place:

• Correctly sequencing the feeds in the load panel using batch run sequencing
• Providing an organized and intuitive feed naming convention for each feed, and a logical workflow

By following these two rules, you empower the end-user and help them better understand their own business processes. This leads to happy customers!