Re: Query execution for intervals
Date: Thu, 9 Feb 2006 15:33:45 +0200
Message-ID: <dsfgbu$oqa$1_at_emma.aioe.org>
<frebe73_at_gmail.com> wrote in message
news:1139489086.727572.103070_at_g44g2000cwa.googlegroups.com...
> If my original assumtion is correct, the response time would be linear.
> The query engine has to traverse all events with the start value lower
> than the given time. In average 50% percent of the events has to be
> traversed. The time could be calculated as:
> time = 0.5 * k * n, where n is the total number of events.
> One possible solution would be that the query engine might use the
> knowledge of the time value compared to the existing start and end
> values, and the average percent ratio might be significant lower than
> 50%. But it would still be a linear response time.
Your k is really a constant or is someting like log(n)/n ?
> > Have you done an experiment to collect some data about the actual
response
> > time for running the query ?
> No, I was hoping to get a qualified answer at this forum.
You can also look at the actual query plan to see what operations would be performed. Received on Thu Feb 09 2006 - 14:33:45 CET