AutoStore: Why Multiple Tasks Are Needed
Our neighborhood market, the big-mega outlets, airport security screening, the DMV, and the interstate freeway system in most major metropolitan areas during rush hour traffic all have one thing in common: there are never enough open lanes.
To the endeavor of order and process, when the volume of activity is not accounted for, regulating the forward flow to a single line creates what we call in the industry as a bottleneck. Stalled movement feels like being tied up in a straight jacket.
One way to think about an AutoStore task how a printer prints. After the print command is launched, bits and bytes received by the printer trigger a series of events. Paper is pulled from the tray, the information is applied to the paper, and the final product is placed on the exit tray.
Before any print job is completed, the printer is in a busy state, and can't be used. Any other jobs that need to be printed are waiting in a queue, and begin after the printer has finished printing the job which preceded it.
A single task created in AutoStore works much like our printer analogy. In single-line fashion, the task must complete the present job before it can take on the next job waiting to be processed. If the task includes processor intensive duties such as OCR, it means that pending jobs must wait.
One of the many reasons to split the effort/work into different tasks in AutoStore is for the very issue described. Each task operates independent of each other, versus the linear task needing to finish the work before consuming the next job in the queue.
For some CFG files, I create a task for capture, and another for various processes. The job gets passed from the capture task into the process/route task using the Knowledge Package Builder (capture task) and the Knowledge Package Loader (process/route task) receiving the job.
Separate capture tasks for multiple capture sources could conceivably now feed separate process tasks, and even separate route tasks, so that a common configuration is administered for each form within the task.
For example, a Xerox, HP, and Kyocera capture task each have 10 forms. If there is OCR needed for each form, then OCR would need to be configured 30 times. A single OCR process task means it would be configured once inside a separate task.
In closing, a single task is fine for a single capture source with relatively low volume. By using multiple tasks, it is possible to scale an AutoStore configuration to handle larger processing by adding more lanes to the workflow.
Cheers!
To the endeavor of order and process, when the volume of activity is not accounted for, regulating the forward flow to a single line creates what we call in the industry as a bottleneck. Stalled movement feels like being tied up in a straight jacket.
One way to think about an AutoStore task how a printer prints. After the print command is launched, bits and bytes received by the printer trigger a series of events. Paper is pulled from the tray, the information is applied to the paper, and the final product is placed on the exit tray.
Before any print job is completed, the printer is in a busy state, and can't be used. Any other jobs that need to be printed are waiting in a queue, and begin after the printer has finished printing the job which preceded it.
A single task created in AutoStore works much like our printer analogy. In single-line fashion, the task must complete the present job before it can take on the next job waiting to be processed. If the task includes processor intensive duties such as OCR, it means that pending jobs must wait.
One of the many reasons to split the effort/work into different tasks in AutoStore is for the very issue described. Each task operates independent of each other, versus the linear task needing to finish the work before consuming the next job in the queue.
For some CFG files, I create a task for capture, and another for various processes. The job gets passed from the capture task into the process/route task using the Knowledge Package Builder (capture task) and the Knowledge Package Loader (process/route task) receiving the job.
Separate capture tasks for multiple capture sources could conceivably now feed separate process tasks, and even separate route tasks, so that a common configuration is administered for each form within the task.
For example, a Xerox, HP, and Kyocera capture task each have 10 forms. If there is OCR needed for each form, then OCR would need to be configured 30 times. A single OCR process task means it would be configured once inside a separate task.
In closing, a single task is fine for a single capture source with relatively low volume. By using multiple tasks, it is possible to scale an AutoStore configuration to handle larger processing by adding more lanes to the workflow.
Cheers!
Comments
Post a Comment