We are developing an ETL process for a source system that has a handful of tables & columns that are also Teradata keywords. We need to preserve the original object names so we have to use double quotes where this is the case.
We can create logic to encapsulate object names when required or we can apply a scattergun approach and encapsulate every object name in double quotes. I prefer the former because I think double quotes are messy and should be avoided where possible but if there is no performance hit with the latter then my personal preference may have to be put to one side for the sake of simpler (or lazy, depending on your point of view!) coding.
Can someone confirm what, if any, impact on performance using double quotes has.
A colleague of mine has told me there is not, so I'll consider this topic closed unless anyone has anything to add.
We are not aware of any performance degradation when using object names in any form (i.e., quoted, unquoted, U&"", or ""XN). Quoted name parsing should be faster as less checking is done, but the difference is negligible within the total processing time. There is a sqlrestrictedwords view which can be queried for names which need to be quoted. Refer to the 14.10 Extended Object Names Orange Book for more details.