Re: Archive sizes

Neil Truby wrote: > >What algorithim does OnLine use when creating archives? I have deleted >a large amount of data from a main table in my application, then >onubload/onloaded the database. I had expected to see the size of the >archive drop substantially, perhaps from 1200MB to 800MB, but in fact >the archive is not very much smaller. I expect that the table now has >a large amount of unused pages allocated, but surely OnLine wouldn't >dump these, would it? Would it?! Neil, the fact that you have deleted the rows on all those pages is also an archivable fact. The algorithm for level-0 is very simple: Back up all occupied pages in all dbspaces. (For OnArhive, is all pages in the specified dbspace set.) An occupied page is one that is allocated to a dbspace. I suspect the timestamp on the page is also part of the decision; something like: If it belongs to a tblspace and its time-stamp indicates something more recent than the creation time of the chunk, then it is occupied. For a level 1 archive, the algorithm is: Back up all pages than have changed since the most recent level-0 arvchive. So even if a page became truly available - eg. you had rebuilt the table with "alter index to cluster" - it would be backed up now because at the time of the last level-0 archive, it was marked differently. The timestamps are used all cases. -- -- Jake (Preserves stuffed packrats) . . _..-'( )`-.._ ./'. '||\\\\. }\\_/{ .//||` .`\\. ./'.|'.'||||\\\\|.. )o o( ..|//||||`.`|.`\\. ./'..|'.|| |||||\\`````` \\'@'/ ''''''/||||| ||.`|..`\\. ./'.||'.|||| ||||||||||||. | .|||||||||||| ||||.`||.`\\. /'|||'.|||||| ||||||||||||{ | }|||||||||||| ||||||.`|||`\\ '.|||'.||||||| ||||||||||||{ | }|||||||||||| |||||||.`|||.` '.||| ||||||||| |/' ``\\||`` | ''||/'' `\\| ||||||||| |||.` |/' \\./' `\\./ \\!|\\ /|!/ \\./' `\\./ `\\| V V V }' `\\ /' `{ V V V \\ \\ \\ V / / / +-----------------------------------------------------------+ | Impeccable Logic: A thought process which successfully | | resists chicken bites | +-----------------------------------------------------------+