![]() | |
![]() |
| | Thread Tools | Display Modes |
#1
| |||
| |||
|
#2
| |||
| |||
|
|
hello, we have a table which will regularly undergo heavy data modifications. this table is not used on remote sites, only on the consolidated. table is not replicated. the modifications flood the log file. do we have any options to somehow manipulate the table with no logging at all? we aware of LOAD TABLE, but the situation is more difficult, updates and deletes are also needed. |
#3
| |||
| |||
|
|
But in any case, the table contents is lost latest when the database is shut down. Depending on your requiremnts, this might or might not be an option. |
#4
| |||
| |||
|
|
On Thu, 15 Nov 2007 12:11:39 +0100, Frank Ploessel fpl... (AT) d_e (DOT) i_m_s_h_e_a_l_t_h.c_o_m> wrote: But in any case, the table contents is lost latest when the database is shut down. Depending on your requiremnts, this might or might not be an option. alas, not. data should be permanent. now we thinking of a separate database and proxy tables, but the performance might be a problem. if all fails, we will use unload table, assemble the new data elsewhere, truncate table and load table to completely replace the data. |
#5
| |||
| |||
|
|
So this implies that the table must be a permanent table - not a temp table. |
|
What transactional semantics would you expect? |
#6
| |||
| |||
|
|
On Thu, 15 Nov 2007 15:48:22 +0100, Mark Culp reply_to_newsgroups_only_please_nosp...here (DOT) com> wrote: So this implies that the table must be a permanent table - not a temp table. that's correct What transactional semantics would you expect? none. the table will be modified by only one specific process, and we require no protection against anything. we can run the data load process at night, on an exclusive connection. we will ensure data integrity ourselves. no need to roll back, no need to recover from failures. inconsistent data is OK. to give a figure: this table is a sales data table which we will receive daily. the table have some thousands of new rows every day, but the new rows are computed based on the old ones and on external data. it may change in the future to some hundreds of thousands each day. this table in fact not part of our database, we don't use it internally, only run some scheduled queries against it joined with other tables that *are* ours, and produce some reports. if we had a separate database for that reason, we would have to transfer all clients, users, products, organization hierarchy, etc to that other database, which is a pain. not to mention many such structures are not stored, but computed by procedures. we don't want to mirror and maintain all these in the sales database. thanks for your effort |
#7
| |||
| |||
|
|
Given these requirements, then external processing to compile the data followed by two steps: - truncate table ... - load table ... |
![]() |
| Thread Tools | |
| Display Modes | |
| |