A transaction is a piece of processing initiated by a single request.
The request is typically made by a user at a terminal. However, it could be made from a web page, from a remote workstation program, or from an application in another CICS® region; or it might be triggered automatically at a predefined time. The CICS web support concepts and structure in in the CICS TS 5.1 Information Center and the Overview of CICS external interfaces in the CICS TS 5.1 Information Center describe different ways of running CICS transactions.
A single transaction consists of one or more application programs that, when run, carry out the processing needed.
However, the term transaction is used in CICS to mean both a single event, and all other transactions of the same type. You describe each transaction type to CICS with a TRANSACTION resource definition. This definition gives the transaction type a name (the transaction identifier, or TRANSID) and tells CICS several things about the work to be done, such as which program to invoke first, and what kind of authentication is required throughout the execution of the transaction.
You run a transaction by submitting its TRANSID to CICS. CICS uses the information recorded in the TRANSACTION definition to establish the correct execution environment, and starts the first program.
The term transaction is now used extensively in the IT industry to describe a unit of recovery, or what is known in CICS as a unit of work. This is typically a complete logical operation that is recoverable; it can be committed or backed out as an entirety as a result of a programmed command or a system failure. In many cases, the scope of a CICS transaction is also a single unit of work, but you should be aware of the difference in meaning when reading CICS documentation.