Metrics: RFC

Christopher Blizzard blizzard at redhat.com
Wed Nov 22 15:39:34 UTC 2006


Mike McGrath wrote:
> So I've compiled thoughts and issues we've come across with our first
> round of metrics in FC6. and created a wiki page so we can actually
> keep track of ideas.  Its a wiki so add/alter stuff as you see fit.
> 
> http://fedoraproject.org/wiki/Infrastructure/Metrics
> 
> I've posed a similar question to the fedora-users list to see how the
> community at large responds.
> 
> https://www.redhat.com/archives/fedora-list/2006-November/msg05080.html
> 

I love how one of the Cons is just "evil."  :)

So those are a lot of possible mechanisms.  Let's talk about goals 
instead?  What do we actually want to know?  Here's what I would love to 
have:

o Rough guess of total number of installs
o Total number of server machines
o Total number of desktop machines
o Number of people who install, use, then never use again
o Total number of people who use on a daily/weekly/monthly basis
o What hardware people are using in the field.

Now I'm going to make an assertion.  I assert that people only hate 
these kinds of things when it gives them no value whatsoever.  That is, 
they don't receive anything in trade for that information.  What we need 
to do is to make sure that in return for becoming part of our network, 
that people feel like they are getting something in return.  Maybe it's 
direct, maybe it's indirect, but still useful.

Let's look at an actual scenario.  I would actually start with the last 
goal on that list to go off of.  If we are able to collect a set of 
hardware profiles for people, just after an install, and tie that to a 
unique machine identifier, we could make that really useful for people. 
  The reason being that having access to information about what hardware 
people are really using allows us to know where we need to concentrate 
our efforts.

DavidZ wrote a little utility for RHEL that basically says "hey, your 
suspend failed, your hardware is busted!" (in friendlier terms.)  What I 
would love to do is to do the same thing with FC6, but with tracking 
over time.  For example, let's say you have a machine and you want to 
suspend it.  That suspend succeeds or fails.  It would be great if both 
success + failures of those were reported to us.  Then we could really 
build a good/bad hardware + driver list.  No one really has this today, 
and it's of clear benefit to our users.

Also, we could include in that data what kernel people were using and if 
we saw a big set of new failures after a certain kernel update, we would 
know that a certain set of hardware was busted in a release.  Really 
useful at a macro level to people like Dave Jones and upstream kernel 
people.

Now, what does this have to do with metrics?  Everything.  Note that the 
above scenario allows us to gather information about desktops, hardware 
and daily users, and has per-machine information, but there's clear 
value to both the people using it and the people providing the software. 
  That's the way that we need to approach gathering metrics.  Give _and_ 
take, not just something we're doing to the people who use the software 
we so lovingly put together.  That's where "Evil" comes from but we want 
to make sure we're not Evil.  Just helpful.

--Chris




More information about the fedora-advisory-board mailing list