Preparing the MAS for history recording

With CICSPlex® SM, you can save and view data for completed tasks; that is, historical task data.

When an active task completes, its data is stored in a historical data store. The data store is made up of a number of VSAM KSDS data sets. You need a minimum of two data sets and a maximum of twenty six data sets.

Each MAS must have its own set of CICSPlex SM history data sets. Allocate the data sets to the MAS region by means of DD cards in the JCL with DD names of the form EYUHISTx, where x is a character suffix taking values A through Z. Dynamic allocation is not supported. Allocate the data sets with a disposition of OLD. Use the suffix letters in ascending sequence with no letters omitted. For example, if four history data sets are required, use DD names EYUHISTA, EYUHISTB, EYUHISTC, and EYUHISTD.

Define the CICSPlex SM history data sets with the REUSE keyword. Task history recording uses the least recently used data set, or, when starting for the first time, EYUHISTA. When EYUHISTA becomes full, it switches to use EYUHISTB and so on in sequence. Each full data set remains open with its data available until the history recorder has filled all data sets and starts reusing the data sets. At this time, EYUHISTA is set closed, emptied, reopened, and reused first, followed by EYUHISTB and so on in sequence. If a data set is reused, its previous contents are destroyed.

Until the history recorder has to empty a data set to reuse it, the historical task data is available for use. The data is maintained across CMAS and MAS restarts. You do not have to define the history data sets as recoverable because unit-of-work recoverability is not required. However, the CICSPlex SM history recorder does require files to be defined as nonrecoverable to avoid unnecessary logging in the MAS region.

In addition, do not define the history data sets to use VSAM compression. The CICSPlex SM history function initializes the data sets to calculate how many records fit in the data set, so that it can safely use sequential writes to the data set, thereby reducing I/O use. Use of VSAM compression spoils that calculation and causes data to be lost when the data set becomes full and a data set switch is required.

CICSPlex SM provides a tuning aid, the HISTRECSMSG EYUPARM parameter, to determine the optimum size for history data sets. HISTRECSMSG can activate the periodic output of messages detailing how many thousand records have been written to the data set. Each completed task has one record. Because CICS® file control supports extended format KSDS data sets, you can define large history data sets over 4 GB in size. However, when considering the use of very large data sets, take into account that when the CICSPlex SM history recorder reuses a data set by emptying it, a large amount of data is lost and not available for subsequent queries. An alternative approach to a small number of very large data sets is to spread the data over more data sets. For example, by having 25 data sets, each one capable of holding one hour's worth of completed task data, at least one day's worth of data can always be maintained. When the oldest data set is reused, only one hour's worth of data is lost.

CICSPlex SM provides a sample job, EYUJHIST, for defining and initializing two history data sets. It is supplied, non-customized, in TDFHINST, and customized by DFHISTAR in XDFHINST.