| abu@accobat.com 2006-03-24, 7:40 am |
| We have migrated a major BI implementation from SQL Server 2000 to SQL 2005.
After migrating the cubes, we are experiencing a major problem with writeback
partitions and the way columns in writetables are named. In 2000, dimension
columns would receive the _L suffix, and measure columns would receive the
SUM_ prefix. In 2005, everything has changed so columns are named with
incremental numbers, e.g. Column_1, Column_2 etc. Unfortunately, the old
writetables are useless with the new cubes. Regardless, all the migrated
writetable partitions have to be
We have also migrated the DTS into the new SSIS format, triggers, and stored
procedures in which extensive amounts of code uses the old naming
conventions. For instance, all entered data from the writetables are
transferred to the fact tables using
Insert into Fact_Finance (money, account)
Select (SUM_money, account_l2) from writetable_Finance
With the new writetables, that code would have to be re-written to:
Insert into Fact_Finance (money, account)
Select (money_0, account_1) from writetable_Finance
In other words, all our existing SQL statements are useless, because they
refer to the old naming conventions.
How can we avoid re-writing all the existing code from _L or SUM_ to using
the new _0, _1, _2 format. The XML file for each cube does not contain the
writeback partion, so it seems impossible to re-edit the writetable column
names back to the old ones, thus avoiding re-writing the code. Is there not
other solution? Can one control or edit how AS2005 names the writetable
columns?
|