Kbase P102141: Performance considerations for CorVu
| Autor |
Progress Software Corporation - Progress |
| Acesso |
Público |
| Publicação |
7/2/2010 |
|
Status: Unverified
GOAL:
Performance considerations for CorVu
FIX:
CorVu performance considerations can be split into those involving data extraction and those involving data manipulation.
Extraction performance is down to the Progress database / database machine, and the efficiency of the SQL generated and then sent to the database. Use the Preview button in Graphical Analysis to check the SELECT statement - examine the SQL from the point of view of the Progress SQL-92 performance.
Extraction performance issues can usually be traced to users placing a constraint on a non-indexed Progress field - e.g. Invoice Date > 1/2/2005, when the invoice date is not indexed in the database table - or joining two tables in the Knowledge Library when the join fields are NOT indexed. The way around these issues is to find alternative joins or constraints, to make a constraint which uses all fields in a Progress table's group index, or to add the relevant index to the database.
For example, if the table index is account code + added-date + line number, then add a constraint where account code > 0, and added-date > 1/2/2005 and line number > 0. This will force Progress to identify the group index and use it for the search but get just the dates required.
If using CorVu with Progress version 9.1D or above, connecting using the Progress SQL-92 driver can be faster, as can using an ABAServer shared memory connection. There are some Progress parameter settings which can help when added to the ABAServer in the Pre-Init Parameters section, e.g. -mmax 1024. Check the Progress documentation for other Progress client startup parameters.
Data manipulation issues depend on the CorVu performance on the machine and operating system - file size, client and server memory, disk i/o (indeed, the usual considerations involved in database tuning), as well as on the complexity of any formula added to the data in the dynamart after data is extracted. Other functions such as summarizing, joining, appending, grouping all take memory - but should work as fast as the machines allow. Using Windows 2000 with 256MB RAM, for example, plus plenty of free disk space, is a good start.