 |
 |
Archives of the TeradataForum
Message Posted: Thu, 05 May 2005 @ 20:40:54 GMT
| Subj: | | Re: Failure 5708 Table header size exceeds limit |
| |
| From: | | Victor Sokovin |
| | I'm getting ready to evaluate PPI on V2R5.1 and this might change some of my test cases (we're reviewing my plan, so I haven't gotten to
the stage of actually running anything). Could this be a problem with CASE_N as opposed to RANGE_N? | |
| | I thought the limit on partitions was 2**16 or is that just the limit on the highest value that can be associated with a partition? | |
I have compared the relevant parts of the "DDL" and "Database Design" manuals for V5.1 and V6.0. It looks like the table header size has
doubled in the latter release, from 64K to 128K. This might explain why the DDL we are discussing here works on V6.0 and does not work on
V5.1.
The maximum number of partitions, whether defined via CASE_N or RANGE_N, is indeed 65535 but this is the theoretical single-parameter limit
that does not take into account any other parameters "sharing" the same 64K/128K worth of space in the table header. It could probably be achieved
(almost) on the simplest one-column table. For the real tables it is not easy to forecast the maximum number of partitions without doing the
calculations. Trying the DDL code on the particular system could be faster but, of course, this method is not applicable in the initial design
phase. Perhaps NCR or other forum subscribers have a (rough) matrix which would facilitate the calculations?
Regards,
Victor
| |
