Guidelines on how to prepare for an anticipated increase in data
Article ID: 33195
Updated On:
Products
Issue/Introduction
There may be situations in which there is an anticipated increase in database size and the amount of processing that change will require. If possible, it is best to plan proactively for this type of change.
Environment
Component:
Resolution
The first thing to do in preparing for an increase in the size of a database is to get as accurate an estimate as possible of the size of the anticipated change. This is likely to be identified as an additional number or percent of record occurrences, for one or more record types. Then use that information to identify the area(s) where the data is stored, and to check them for space availability. The second-tier things to check are:
1- Index efficiency;
2- Records that are stored via; and
3- Other physical factors that could influence performance.
The best place to start this assessment is with a general review of statistics. You can see those by using the DCMT DIS STAT SYS command, checking for the following:
1- On a typical day, do you see any deadlocks, or does the system hit MAX TASKS?
2- Do you see any CALC or VIA overflows? Or Buffer waits?
3- Under indexing statistics, do you see any splits or spawns?
Any of the above could mean that your database needs to be tuned. You should remedy any existing problems first, then repeat these steps after the new records are loaded to ensure that the additional records have not compromised your system efficiency.
You may want to also use various reports produced by IDMS Performance Monitor (PERFMON) to assess the current efficiency of the database functioning. Detailed instructions on how to do this are listed in the knowledge documents referenced below.
If your system relies heavily on indexes to access important records efficiently, you should use the PRINT INDEX utility to check the indices defined on those records. The output can give you an idea of how efficiently the indexes are structured. Thinking of the anticipated increase in the number of record occurrences, it is prudent to run some calculations to see if the IBC count of those indexes should be changed to accommodate this increase. If the calculations indicate that a change is warranted, the indices should be rebuilt with that IBC count, allowing free spaces in the indexes, before loading the new records. Complete instructions on how to calculate the most efficient values for the index parameters, see the manual noted below.
Additional Information:
Knowledge document TEC571497 contains details on how to diagnose database application performance issues.
Knowledge Document TEC543787 contains details on how to do a thorough database performance analysis
Detailed information about how to calculate the most efficient structure for an index can be found in the IDMS Database Design Guide, in the section titled “Calculating the Size of the Index”. The manual for 19.0 can be found here:
Feedback
