… since I posted here. In part I have been delegated to gardening duties, but as the need for muscle in the garden is giving way to artistic flair I can revert to my shadowy internet existence and write on whatever takes my fancy…
Which will probably be a bit more on summary design practicalities – for example when are indexes better than summary tables? Perhaps I might also write a bit on a project I am managing to upgrade OWB from 9.2 to 10.2 for one of our customers (or, if you are in London next month you could go and hear Jon Mead giving a similar talk at the UKOUG BIRT SIG) in fact the whole project is to move the customer’s reporting tool from Windows to Solaris and upgrade the database from 9i to 10g so there is scope for other articles there too.
Other interesting topics might include “why doesn’t my query eliminate partitions?” but I will have to work out the answer on that one first; suffice it to say the CBO is a little too fond of a nested loop join with the whole of a 800+ partition table. I can hint the query to make it work, but hinting is never the answer – it is just a tool to identify where the statistics collection is sub-optimal.
I have also been talking to our content management guys and that has lead me to think about indexing, and not just for text. Electronic content storage takes massive amounts of storage but unlike data warehousing we are only seeking single records (or perhaps just a few that match a predicate.) Here efficient indexing is very important to find matches in a reasonable amount of time. Which in turn got me thinking about spatial, and spatial on the small scale; say using it to compare the polygons defined by the vertices between eyes, nose, mouth and chin and using it to identify people
… however as the garden will be open to the public in July I may have get those green fingers out again and stop writing again.