But i am having a speed problem with the dataimport, when the dataimport
gets to the table LedgerJournalTrans, the import slows down quite a lot.
Before i get to that table, it says that dataimport will take 4 hours,
but after having been importing records into that table for now more
that 10 hours, it now says 38 hours.
I have to import 117503 into LedgerJournalTrans and it has now gotten to
104072 records, and it seems to be getting slower and slower.
Does anyone have an idea as to what is slowing it down this much?
RecId recalculation mechanism. Standard data import is too slow for big
amounts of data. Almost unreal.
But 100,000 records is not so big amount to be unreal. Just very big.
Ok so any idea what can be done to make it faster?
I am not sure if that was good idea, but I wrote special data import
procedure to import entire company (duplicated standard class of data import
and wiped a lot of code). The idea was to import data into clean company and
do not recalculate RecId and SequenceId. Indicators showed little bit more
than 1000 records per second import speed. It is above 3,600,000 records per
hour. But due to fact there was no RecId recalculations import allowed to
divide all tables into different data definitions groups with almost equal
number of records and run import from different clients simultaneously
increasing speed in several times.
After such an import I run RecId compression procedure (also customized
standard procedure; customized because standard procedure has serious
performance problems).
My task also was moving data from Oracle to MS SQL.
All that worked in 3.0. In 4.0 that code was not operational because of
changes in system classes. Someone from this group asked me to help him to
realize similar idea for 4.0, but unfortunately I did not have time that
time to help and still do not have code for later versions.
If you will not find any other solution you may try this one.
Oh that sounds nice, i would like to get that solution.