Archives of the TeradataForum

Message Posted: Fri, 02 Feb 2007 @ 10:06:43 GMT

	<Prev	Next>		<<First	<Prev	Next>	Last>>

A couple comments on John Hall's strawman...

- Why compress small tables since there is much less impact to total space? I agree that the big tables should be done first to get the maximum payback. However, small tables with a very high frequency of access benefit from being able to be cached in memory. If they are smaller, they are more likely to fit and more of them will fit.

- Re "saving nodes": Saving nodes cannot be calculated only on space. Both space and compute/IO resources must be considered. MVC's space savings is not in the same proportion as the resource savings. Thus, purely for example your mileage will vary, if you save 30% of space you may well only save 5% of CPU/IO for a particular workload. Thus the number of nodes saved would only be 5% if holding response time/throughput constant was the goal.

And a general comment on this thread: In addition to the cost savings on the system, the cost of your time and energy to build maintain and operate your own tools/scripts should be considered. If a tool can do the job more quickly/easily and you can do other things with your time, there is value to that.

	<Prev	Next>		<<First	<Prev	Next>	Last>>