Comments on: Tuning 101, Part 1: Egads! Why is it So Slow? http://www.pythian.com/news/235/tuning-101-part-1-egads-why-is-it-so-slow/ News and views from Pythian DBAs Fri, 10 Feb 2012 13:01:25 +0000 hourly 1 http://wordpress.org/?v=3.0.4 By: Good Database Design is Mightier than Hardware http://www.pythian.com/news/235/tuning-101-part-1-egads-why-is-it-so-slow/#comment-168812 Good Database Design is Mightier than Hardware Fri, 07 Mar 2008 15:03:14 +0000 http://www.pythian.com/blogs/235/tuning-101-part-1-egads-why-is-it-so-slow#comment-168812 [...] one of my previous blog posts, I mentioned something along the lines of—well I’ll just cut and paste . . . In my [...] [...] one of my previous blog posts, I mentioned something along the lines of—well I’ll just cut and paste . . . In my [...]

]]> By: zfriese http://www.pythian.com/news/235/tuning-101-part-1-egads-why-is-it-so-slow/#comment-140079 zfriese Wed, 12 Dec 2007 14:34:57 +0000 http://www.pythian.com/blogs/235/tuning-101-part-1-egads-why-is-it-so-slow#comment-140079 "It takes unbound SQL and strips away the bind variables." Am I missing something, or does this routine actually strip away literals, rather than bind variables? “It takes unbound SQL and strips away the bind variables.”

Am I missing something, or does this routine actually strip away literals, rather than bind variables?

]]> By: joel garry http://www.pythian.com/news/235/tuning-101-part-1-egads-why-is-it-so-slow/#comment-1941 joel garry Thu, 31 Aug 2006 00:29:55 +0000 http://www.pythian.com/blogs/235/tuning-101-part-1-egads-why-is-it-so-slow#comment-1941 "In the hands of an experienced tuner, the cost of getting more out of the whole system does not compare with the cost of a hardware upgrade. It’s just apples and oranges… and my apples are much cheaper than their oranges." This breaks down when the hardware is approximately 3 generations old, where generation is defined as the doubling of compute power per unit cost (commonly considered to be about 18 months). Now, if the experienced tuner only works on new systems where "I upgraded to the new version of Oracle and now it runs slower!" a tuner may be your sort of apple. In my orchard, systems running badly for years before I see them is pretty common. So the apples and oranges really means non-exclusivity - you can fix some stuff (tuning) before making a more global fix (capital expenditure). And doing it the other way around (or preferably, iteratively) can make for happy customers. Let's see... Apples are $60/hour and Oranges are $40,000... but a happy customer will buy both! “In the hands of an experienced tuner, the cost of getting more out of the whole system does not compare with the cost of a hardware upgrade. It’s just apples and oranges… and my apples are much cheaper than their oranges.”

This breaks down when the hardware is approximately 3 generations old, where generation is defined as the doubling of compute power per unit cost (commonly considered to be about 18 months). Now, if the experienced tuner only works on new systems where “I upgraded to the new version of Oracle and now it runs slower!” a tuner may be your sort of apple. In my orchard, systems running badly for years before I see them is pretty common. So the apples and oranges really means non-exclusivity – you can fix some stuff (tuning) before making a more global fix (capital expenditure). And doing it the other way around (or preferably, iteratively) can make for happy customers.

Let’s see… Apples are $60/hour and Oranges are $40,000… but a happy customer will buy both!

]]>