I have a table with 900000 records. ( File size is 1.3 GB )
The table is indexed on ID and NAME. ( there are other fields )

INDEX ON ID TAG ID

INDEX ON NAME TAG NAME

In VFP7 both are very fast:
(I want to fins all records where the name field starts with DEKK)

SELECT ID,NAME FROM <TABLE> WHERE NAME = 'DEKK'

SELECT ID,NAME FROM <TABLE> WHERE NAME LIKE 'DEKK%'

410 record are selected in no time in both cases.

When I use ADODB.RecordSet and VFPOLEDB:

SELECT ID,NAME FROM <TABLE> WHERE NAME = 'DEKK'

does not work (after 370
seconds no records are selected)
Therefore I have to use:

SELECT ID,NAME FROM <TABLE> WHERE NAME LIKE 'DEKK%'

This also takes forever. ( 360 - 380 seconds ) recordset.recordcount is 410.

SELECT ID,NAME,DESC FROM <TABLE> WHERE DESC LIKE 'WINTER%'

also takes 360 - 380 seconds. This query is on a field without index.

To me it seems like VFPOLEBD is ignoring the indexes.
Isn't VFPOLEDB supposed to used indexes?
Does VFPOLEDB need more memory to work well?
Or is it designed to be slow on large tables?
What performance can I expect from MSDE and SQL Server?

(Old FoxPro 2.6 for DOS does the job in 330 seconds, so there is no need to
complain about VFP7 doing it in 0.1 seconds)