6. Database Tailoring and Retrofitting Techniques › 6.2 Tailoring the Database › 6.2.2 Tailoring Database Files › 6.2.2.3 Change a File's Sequence in a Timespan
6.2.2.3 Change a File's Sequence in a Timespan
The most complicated level of file tailoring involves
changing data element statements to modify the sequence of
observations within a file.
The statements listed below define an example Network
Performance Analyzer (NPA) component with one file, the
Network Control Program File. These statements would be
found in the components cccGENIN member in
sharedprefix.MICS.GENLIB. Data element definition statements
that define the sequence of file observations in any timespan
are indicated by "-->" in the left margin of the example.
*
* NPA component generator statements
*
GEN GENFILES
COMP NPA 050 32000 VBS NOACCT NPA Component
AREA NPA NPA Activity Information Area
FILE NCP 00 1 Y Y Y Y N N Y Y NCP Activity File
TYPE R 3 . 3 . 3 .
NAME NCPNRFBQ 00 0 0 0 0 0 Free Buffer Queue Length
NAME NCPNRHQL 00 0 0 0 0 0 NCP Channel Hold Queue Length
NAME NCPNRIQL 00 0 0 0 0 0 NCP Channel Intrmd Queue Len.
NAME NCPNRSFB 00 0 0 0 0 0 Free Buffer Count At Slowdown
TYPE A 5 TIME11.2 6 TIME15.2 6 TIME15.2
NAME NCPTMFCT 00 0 0 0 0 0 Free Cycle Time
NAME NCPTMTIS 00 0 0 0 0 0 Time In Slowdown
TYPE M 3 . 3 . 3 .
NAME NCPMXANP 00 0 0 0 0 0 Max Available NCP Buffers
NAME NCPMXFBH 00 0 0 0 0 0 Free Buffer High Water Mark
TYPE N 3 . 3 . 3 .
NAME NCPMNFBL 00 0 0 0 0 0 Free Buffer Low Water Mark
NAME NPANRRLS 99 0 0 0 0 0 .
NAME NPANRSLS 99 0 0 0 0 0 .
NAME NPATMAOT 99 0 0 0 0 0 .
NAME NPATSEIS 99 0 0 0 0 0 .
NAME NPATSFRU 99 0 0 0 0 0 .
NAME NPATSLOG 99 0 0 0 0 0 .
NAME RESNAME 99 0 0 0 0 0 .
NAME MICSVER 99 0 0 0 0 0 .
-->NAME DAY 99 5 5 N N N .
-->NAME HOUR 99 6 6 6 N N .
-->NAME WEEK 99 0 0 4 N N .
-->NAME YEAR 99 3 3 3 3 3 .
-->NAME MONTH 99 4 4 N 4 N .
NAME INTERVLS 99 0 0 0 0 0 .
NAME DURATION 99 0 0 0 0 0 .
NAME STARTTS 99 0 0 0 0 0 .
-->NAME ENDTS 99 7 0 0 0 0 .
-->NAME ZONE 99 0 0 5 5 4 .
NAME DAYNAME 99 0 0 N N N .
-->NAME SYSID 99 1 1 1 1 1 .
-->NAME NCPNAME 99 2 2 2 2 2 .
NAME NETNAME 99 0 0 0 0 0 .
Note the timespan indicators. For data elements that
merely exist on the file in all timespans, the indicators
are of the form:
NAME NETNAME 99 0 0 0 0 0 .
---------
indicators
Data elements that are part of the key to the file in any
timespan have timespan indicators that are numbers other
than zero in one or more timespans. Consider ZONE:
NAME ZONE 99 0 0 5 5 4 .
| | | | |
| | | | +- 4th item in key to YEARS
| | | +--- 5th item in key to MONTHS
| | +----- 5th item in key to WEEKS
| +------- exists, not in key to DAYS
+--------- exists, not in key to DETAIL
The NAME statements which define sequence data elements
determine how observations are arranged in the timespans
shown. This table gives the data elements used as keys to
the sort operations that order the file:
DETAIL DAYS WEEKS MONTHS YEARS
------- ------- ------- ------- -------
SYSID SYSID SYSID SYSID SYSID
NCPNAME NCPNAME NCPNAME NCPNAME NCPNAME
YEAR YEAR YEAR YEAR YEAR
MONTH MONTH WEEK MONTH ZONE
DAY DAY ZONE ZONE
HOUR HOUR HOUR
ENDTS
Note that this file does not exist in the YEARS timespan,
according to the FILE statement. If it did, the order
would be as shown, due to coding in the sequence numbers in
the timespan indicators of the NAME statements for the key
data elements.
Suppose the desired order included another data element.
In this example, let us add NETNAME to the order. We will
add NETNAME immediately after NCPNAME in all timespans, to
yield:
DETAIL DAYS WEEKS MONTHS YEARS
------- ------- ------- ------- -------
SYSID SYSID SYSID SYSID SYSID
NCPNAME NCPNAME NCPNAME NCPNAME NCPNAME
NETNAME NETNAME NETNAME NETNAME NETNAME
YEAR YEAR YEAR YEAR YEAR
MONTH MONTH WEEK MONTH ZONE
DAY DAY ZONE ZONE
HOUR HOUR HOUR
ENDTS
Such an order can be constructed by coding the timespan
indicators as:
NAME RESNAME 99 0 0 0 0 0 .
NAME MICSVER 99 0 0 0 0 0 .
NAME DAY 99 6 6 N N N .<--+
NAME HOUR 99 7 7 7 N N .<--+
NAME WEEK 99 0 0 5 N N .<--+
NAME YEAR 99 4 4 4 4 4 .<--+
NAME MONTH 99 5 5 N 5 N .<--+- data elements modified
NAME INTERVLS 99 0 0 0 0 0 . |
NAME DURATION 99 0 0 0 0 0 . |
NAME STARTTS 99 0 0 0 0 0 . |
NAME ENDTS 99 8 0 0 0 0 .<--+
NAME ZONE 99 0 0 6 6 5 .<--+
NAME DAYNAME 99 0 0 N N N .
NAME SYSID 99 1 1 1 1 1 .
NAME NCPNAME 99 2 2 2 2 2 .
+->NAME NETNAME 99 3 3 3 3 3 .
|
+- data element added to sequence
On the surface, this is a very simple change. It is, for
the new user who has been fortunate enough to make such a
modification before the initial installation of CA MICS.
For users that have installed CA MICS and wish to preserve
the existing data, problems arise.
The first problem is that timespan cycles already closed
have been sequenced in an order different from the new key.
To merge yesterday's data with today's data, for example,
would require special processing. Such problems occur in the
DETAIL and DAYS timespans of the online CA MICS database,
and with the Audit Archive files associated, if active.
More importantly, existing to-date files in WEEKS,
MONTHS, and YEARS timespans have already been sequenced in
the previous order. Data following the sequence change
cannot be merged with month-to-date data, for example,
because the keys are unlike. SAS will terminate the daily
run to update the component that contains such a file if the
to-date files have not been retrofitted with the new key
structure.
This problem is almost identical to the problem of adding or
reconfiguring account codes. See Section 6.3, "Retrofitting
Guidelines," for information on how to retrofit a file to
accommodate a modified file sequence.
****************************
* Making the Modifications *
****************************
To modify the sequence of a file in one or more timespans,
perform the following:
- Modify the MCG NAME statements for the data elements
needed to modify the file's sequence.
- If such a retrofit is needed, follow the instructions
in Section 6.3, "Retrofitting Guidelines," of this
document.
- Run the component generation for the component in
question.
- The next daily update for any database unit in which
the component has been installed will reflect the
change, and add the data element to the file in the
desired timespan(s).
