The idea is to unload lobs just from the lob segment.
We need to find out:
1. Is it a clob or a blob
2. for clobs the character set, single byte or UCS16, byte size
3. the chunk size, lob page# in the header is a fat page number
4. missing pages should be emitted as all zeroes
5. size is unknown, simply strip off trailing zeroes (but respect byte size)?
Changes implemented:
1. Add lob segment id to scanned lob page info as most properties apply to the segment. that is the layout of the scanned lob pages cache is now: segid, lobid, fatpageno (chunk#), version wrap, vsn base, ts#, file#, block#
Things to do:
1. Command to unload all lobs a lob segment provide all common properties in the command
2. Command to unload a single lob from lob segment specifying lobid, optional size, and optional dba list
3. Analyze command to produce command to unload lob segment
4. Analyze command to produce commands to unload each lob from segment
Things to consider:
1. change file#, block# to a single dba? pro no calculations, contra more difficult to read?
Leave a Reply