User guide

Improving Performance
Using ENFORM Efficiently
058058 Tandem Computers Incorporated 5–5
STRATEGY COST = 7
The query processor might have to sort two files more than once for a link. A
strategy cost of 7 might occur when:
A query involves at least three files.
The query processor uses a join strategy on the last two files read.
The query contains a request qualification that compares fields from both of
the files involved in the join strategy with fields from a nonjoin file.
The query processor must sort both files involved in the join strategy.
STRATEGY COST = 8
The query processor reads more than one file completely at least once with no
convenient strategy for linking the record descriptions. A strategy cost of 8 might
occur when two or more files are linked with non-equality comparisons in a
request qualification. A strategy cost of 8 frequently occurs when a query links
two or more FIND files in which the data is stored in the default unstructured file
type.
Improving
Performance
The following paragraphs suggest changes that improve the performance of ENFORM
queries. Consider making changes in three areas:
The data (disc) environment
The nondisc environment
The wording of the query itself
Only make changes that: are easy to implement, generate significant reduction in
query response time, and increase transaction rate (the updating and retrieval of data)
while reducing response time.
Several of the suggested changes are helpful in all circumstances. Others are helpful
only for particular cases. Use your knowledge of your own environment to determine
the changes most helpful to you.
Changing the Data (Disc)
Environment
The most desirable changes to the disc environment both decrease response time and
increase the transaction rate. These changes involve reducing the number of physical
file accesses by:
Removing levels of indexing in key-sequenced files.
Adding or removing alternate keys.
Avoiding sorting of an already sorted file.
Specifying where ENFORM builds temporary work files.
Spreading input/output demands among discs.
Altering cache size.
Controlling the size of the target file.