Also, there are plenty of Palomino tools around: https://github.com/palominodb/PalominoDB-Public-Code-Repository/ these are in use many places, much wider than the spread of just PalominoDB (For example the nagios plugin is distributed by the Nagios corporate folks). Is there a way to “Submit” a tool? I like the idea of having one place where you can see them, but this seems to be mostly hackmysql tools + openark + percona tools.

As you noted, there’s a problem categorizing things like Workbench: is it a tool or a suite of tools? It doesn’t seem to be a tool, but if it’s the latter, then how does that fit the hierarchy? Such issues probably arises only rarely in the natural world. A griffin comes to mind: part lion, part eagle. Workbench is decidedly less mythical, though. :-)

As for Wikipedia, that’s a good idea, although I’ve never created a Wikipedia page before. Perhaps I’ll try.

Speaking for my own (sadly not represented) team who develops a “Monitor Tool”, “Analyzer Tool” and “Detector Tool” in one bundle, I wonder a little about the hierarchy that you have as well – surely an “Advisor” tool needs to sit under a “Monitor”, with “Detector” and “Advisor” being the same class of thing (a Detector tool still has “a certain amount of human knowledge which they pair with the target condition” – because a human writes the detector, that’s not a good distinction between the two, they do the same thing, they are both written by humans).

Wikipedia has something like an attempt at this, but I think your approach servers that overview much better:

http://en.wikipedia.org/wiki/Comparison_of_database_tools

It would be great to see your approach translated in to something on Wikipedia, that could be added to by all..

10k @ 1s
1k @ 5s
10k @ 1s

So the 5s spike values are exactly 5% of total values (1k/20k). That make the 95%/5% edge something like 1s/5s. But the 5s values aren’t all 5s, some are 2, 3, or 4. So the edge is more like 1s/2s. Then (I should have probably made this explicit in the post), 95% here is actually “95% of values less than or equal to X” whereas technically a percentile should be just “less than X”. Using <= is inherited from mk-query-digest, but perhaps I should re-examine if that's correct.

wow – this is great, very interesting stuff :)

I’m now downloading your code to check out the algorithm – I’m really curious how it works.

Oh, to get a quick grasp of the result, I calculated the correlation between (REAL_95,OLD_95) and (REAL_95,NEW_95), and the results are 0,9998926355 and 0,9999572588 respectively – I think it’s safe to say both are accurate enough, and it seels the new one is even a bit more accurate than the old one (for this dataset).

Exciting :)