Archives of the TeradataForum
Message Posted: Tue, 25 Sep 2001 @ 11:17:40 GMT
| Subj: || || ROWID as UPI of "Key" Table |
| From: || || Claybourne Barrineau |
I'm experimenting with the suggestion of using the ROWID of my Atomic Fact table as a UPI of a "transparent key" table for a Star Schema
design. For some reason, the Fact table is being redistributed to all Amps before this "key" table is built. Shouldn't these ROWIDs exist
on the same AMPs as the Primary Index of the Atomic Fact table? Therefore, a redistribution shouldn't be required.
Insert into INFVO2PLYT.T_F_VOL_KEY_NON_NUPI
| ||1)||First, we lock a distinct INFVO2PLYT."pseudo table" for write on a RowHash to prevent global deadlock for
| ||2)||Next, we lock a distinct INFVO2PLYT."pseudo table" for read on a RowHash to prevent global deadlock for
| ||3)||We lock INFVO2PLYT.T_F_VOL_KEY_NON_NUPI for write, and we lock INFVO2PLYT.T_F_VOL_DY_AC_PS_CB_ST_B4 for read.
| ||4)||We do an all-AMPs RETRIEVE step from INFVO2PLYT.T_F_VOL_DY_AC_PS_CB_ST_B4 by way of an all-rows scan with no residual conditions into
Spool 1, which is redistributed by hash code to all AMPs. Then we do a SORT to order Spool 1 by row hash. The input table will not be
cached in memory, but it is eligible for synchronized scanning. The result spool file will not be cached in memory. The size of Spool 1 is
estimated with high confidence to be 8,000,715 rows. The estimated time for this step is 2 hours and 43 minutes.
| ||5)||We do a MERGE into INFVO2PLYT.T_F_VOL_KEY_NON_NUPI from Spool 1 (Last Use).
| ||6)||We spoil the parser's dictionary cache for the table.
| ||7)||Finally, we send out an END TRANSACTION step to all AMPs involved in processing the request.
| ||->||No rows are returned to the user as the result of statement 1.