Liveware Publishing Support Forum

We have a client who has a large database. The tables are currently stored in an Xbase format and R&R takes a very long time to read and select the records using linked tables. We have considered running a mirror of the Xbase files in sql and using R&R Sql to run the reports. From our tests it appears that we aren^t gaining a whole lot of speed because it is not passing the query to sql and letting sql do the joins; R&R appears to want to read the records on its own, therefore not gaining any more speed. Are we correct in this or is there some way to gain more speed when running reports against a large database?

using RR-SQL does not increase performance unless you move the data to a REAL backend database (like SQL Server, Oracle) - __a mirror of DBF files is just another set of DBF files and i would suggest that the normal xBase version of RR would actually outrun using RR-Sql and an DBF ODBC driver - you have probably introducing more code and it would actually run slower with RR-SQL.____if you stay with DBF files - you need to do some serious design work on the indexes used with the DBFs - that is where you can gain significant performance - anything that avoids "end-to-end" searching will speed up your RR reporting.____if you can afford a REAL backend database - you will notice faster processing simply because you are not bringing the entire DBF file down to your workstation just to select a few records - ____so you have 2 choices - redesign the mirror DBFs with some good indexes and use them with xBase RR -__or invest in SQL SERVER or ORACLE and let it process the large database on the separate server.____i have used RR-SQL against both SQL SERVER and ORACLE and the speedup is entirely based on the loading of the backend database .__I have also loaded very large DBF files (millions of rows) - but added index files which speeds up the processing - as the index files are loaded into memory and processed first - then only the records needed are retrieved from the DBF files.____i hope this helps - __nick__