it's not realy a fragmentation but ..??
Hi, something seams strenge to me. Ihave created a table in one dbspace, and an index in an ather dbspace(( CREATE TABLE ... IN dbspace1); (CREATE INDEX ... IN dbspace2)). normaly it's a simple detached index, but informix insert a entry in the sysfragments tables indicating that the index is fragmented. I find that not normal and specialy when I do that with dbspace1=dbspace2 (it insert a entry in sysfragnets to). Please do you find that normal and if yes can you give me an explication. thinks. ______________________________________________________ Get Your Private, Free Email at http://www.hotmail.com
Since 7.21 Informix treats all detached indexes, and all attached indexes on fragmented tables as well, as fragmented indexes with only one fragment, strategy of "I" for Index or "T" for "Table tracking", respectively. I think that they would like to treat all tables the same way for consistency and do simplify the server code but there is just too much legacy software out here that expects to see a non-zero partnum for non-fragmented tables. Art S. Kagel dragon dragon wrote: > > Hi, > something seams strenge to me. Ihave created a table in one dbspace, and an > index in an ather dbspace(( CREATE TABLE ... IN dbspace1); (CREATE INDEX ... > IN dbspace2)). normaly it's a simple detached index, but informix insert a > entry in the sysfragments tables indicating that the index is fragmented. I > find that not normal and specialy when I do that with dbspace1=dbspace2 (it > insert a entry in sysfragnets to). > Please do you find that normal and if yes can you give me an explication. > thinks. > > ______________________________________________________ > Get Your Private, Free Email at http://www.hotmail.com
OK an attached index on a fragmented table would have the same fragmentation as the table, hence the strategy flag 'T'. Sorry, my language got nested a bit more deeply than usual and even I lost track. Art "Art S. Kagel" wrote: > > Since 7.21 Informix treats all detached indexes, and all attached > indexes on fragmented tables as well, as fragmented indexes with only > one fragment, strategy of "I" for Index or "T" for "Table tracking", > respectively. I think that they would like to treat all tables the > same way for consistency and do simplify the server code but there is > just too much legacy software out here that expects to see a non-zero > partnum for non-fragmented tables. > > Art S. Kagel > > dragon dragon wrote: > > > > Hi, > > something seams strenge to me. Ihave created a table in one dbspace, and an > > index in an ather dbspace(( CREATE TABLE ... IN dbspace1); (CREATE INDEX ... > > IN dbspace2)). normaly it's a simple detached index, but informix insert a > > entry in the sysfragments tables indicating that the index is fragmented. I > > find that not normal and specialy when I do that with dbspace1=dbspace2 (it > > insert a entry in sysfragnets to). > > Please do you find that normal and if yes can you give me an explication. > > thinks. > > > > ______________________________________________________ > > Get Your Private, Free Email at http://www.hotmail.com