Question 27

Explanation

The main improvement of transactional processing over the iterative processing is the ability to get and process the data are completely independent.

https://forum.uipath.com/t/difference-between-iterative-process-and-transaction-process/288748/2

Transaction Processing

What is a transaction?

A transaction represents the minimum (atomic) amount of data and the necessary steps required to process the data, as to fulfill a section of a business process. A typical example would be a process that reads a single email from a mailbox and extracts data from it.

We call the data atomic because once it is processed, the assumption is that we no longer need it going forward with the business process.

Considering the steps of a business process and how they are repeated, we can divide business processes into three categories:

Linear

The steps of the process are performed only once and, if there is the need to process different data, the automation needs to be executed again. For example, if we go back to the email example from this chapter's introduction, and a new email arrives, the automation needs to be executed again in order to process it.

Linear processes are usually simple and easy to implement, but not very suitable to situations that require repetitions of steps using different data.

Iterative

The steps of the process are performed multiple times, but each time different data items are used. For example, instead of reading a single email on each execution, the automation can retrieve multiple emails and iterate through them doing the same steps.

This kind of process can be implemented with a simple loop, but it has the disadvantage that, if a problem happens when processing one item, the whole process is interrupted and the other items remain unprocessed.

Transactional

Similarly to iterative processes, the steps of transactional processes repeat multiple times over different data items. However, the automation is designed so that each repeatable part is processed independently.

These repeatable parts are then called transactions. Transactions are independent from each other, because they do not share any data or have any particular order to be processed.

The three categories of processes can be seen as maturity stages of an automation project, starting with simple linear tasks, which then are repeated multiple times and finally evolve into a transactional approach.

However, this is not a absolute rule for all cases, and the category should be chosen according to the characteristics of the process (e.g., data being processed and frequency of repetitions) and other relevant requirements (e.g., ease of use and robustness).

Business scenarios for transaction processing

• You need to read data from several invoices that are in a folder and input that data into another system. Each invoice can be seen as a transaction, because there is a repetitive process for each of them (i.e. extract data and input somewhere else).

• There is a list of people and their email addresses in a spreadsheet, and an email needs to be sent to each of them along with a personalized message. The steps in this process (i.e., get data from spreadsheet, create personalized message and send email) are the same for each person, so each row in the spreadsheet can be considered a transaction.

• When looking for a new apartment, a robot can be used to make a search according to some criteria and, for each result of the search, the robot extracts the information about the property and insert the data into a spreadsheet. In this case, the details page for each property constitutes a transaction.