I've wanted to do this for several years , the project is
dbcollation.org
I actually proposed for UC but it was objected.
Happy to share my vision and work plan if enough people are interested.
Ronald
Sent from my iPhone
On Dec 25, 2008, at 22:31, Ian Holsman <[email protected]> wrote:
Brian Aker wrote:
Hi!
One of things we have thought about doing for years is a phone home
project. Namely... have the database, via opt-in, report back some
usage information. This would be for telling us what sort of
hardware we are being run on, engines used, etc...
What do people think about this?
A few first order rules ->
1) All reporting information is completely public.
2) We are opt-in, you have turn this on for it transmit information
What sorts of information would people be ok to share/want to see?
First. I think the information will be skewed toward developer boxes
and smaller installs, as people will probably say 'yes' when they
install it on their workstations, but not when they install it on a
50-node farm, as those installs would be more automated in a
traditional shop, and in my experience operations people don't like
having things running on production machines that aren't absolutely
necessary.
secondly you would need to have it phone home on a regular basis,
especially if you want to compare configs, or compare sizes, this
would make it less likely to get it run on a production box I would
think.
thirdly. I'd say I would want this information at the cluster level.
most people don't have a single mysql/drizzle instance. It would be
handy to have stats/info at the cluster level.
as for what info I would like to see.
* install type. 'new user/newbie', 'developer', 'default/un tuned
production install', 'tuned/production install'. This help others to
see if they should be looking at the config in the first place. (and
to help differentiate the unwashed masses and the people who have
spent some time tuning their install)
* the results of a quick/standard set of benchmarks representing how
people would typically use it. (no idea on how to accomplish this)
* the output of information_schema.
* the physical config of the drizzle cluster and how it the machines
are configured together.
Regards
Ian
Here is an example of what the Fedora project does with its phone
home data:
http://smolts.org/static/stats/stats.html
Cheers,
-Brian
--
_______________________________________________________
Brian "Krow" Aker, brian at tangent.org
Seattle, Washington
http://krow.net/ <-- Me
http://tangent.org/ <-- Software
_______________________________________________________
You can't grep a dead tree.
_______________________________________________
Mailing list: https://launchpad.net/~drizzle-discuss
Post to : [email protected]
Unsubscribe : https://launchpad.net/~drizzle-discuss
More help : https://help.launchpad.net/ListHelp
_______________________________________________
Mailing list: https://launchpad.net/~drizzle-discuss
Post to : [email protected]
Unsubscribe : https://launchpad.net/~drizzle-discuss
More help : https://help.launchpad.net/ListHelp
