End of MySQL?

[richb-RBmg6HWzfGThzJAekONQAQ at public.gmane.org (Rich Braun)]

"Mark J. Dulcey" <mark-OGhnF3Lt4opAfugRpC6u6w at public.gmane.org> wrote:
> The catch is that increasingly MySQL DOES compete with Oracle.
> ...for large enterprises and improved performance for large databases.
> ... not so much that existing large companies switch, but that
> small companies stay with MySQL as they grow ...

> I would expect Oracle to chop off that trend off at the knees

Of the various comments posted since I launched this thread, I'm in closest
agreement with Mark's statement here.

I'm at one of those small companies that's grown to the point where we've
bought a couple of Oracle servers and pondered the question whether to switch.
 For us it would mean a *fundamental* change in how we view databases.  Right
now we spawn MySQL instances the way a typical VMware-based shop spawns
Windows instances.  Exceedingly useful for developers, QA, and systems
maintainers.

Go with Oracle, and all that comes to an abrupt halt.  First we have to cut a
check for about $2 million.  Then we have to eBay about 50 used computers that
we can no longer run anymore because it would cost too much to put Oracle on
them.  And we have to tell QA to live with less, and tell the DBA team to be a
lot more careful with the databases because there's no more sandbox to play
in, everything has to go through a disciplined big-company process.

There is a mindset that goes along with Oracle and its enterprise-grade rivals
that is just *so* different from the average MySQL shop.

As for the question whether MySQL has "grown up" enough to compete with
Oracle:  the answer would have been "absolutely!" if Sun hadn't bought out the
operation 3 months prior to the scheduled 5.1 release.  Loss of key
developers, a delay in the schedule, and a number of botched internal
processes in the QA and release process meant that 5.1 really isn't an
effective competitor for some of the features that my company is currently
using.  Having recently hired a couple of combination Oracle-MysQL DBAs, I'm
faced with a practically-daily reminder about how badly MySQL sucks compared
to Oracle on various things that we currently put up with.  Examining the
specifics of each instance of suckitude, I can put the blame squarely on the
fact that 5.1 wasn't really finished before Sun heaved it out to the
unsuspecting public.  (Fortunately, we're very suspicious and we've gotten
pretty good at developing workarounds...)

Net result for us is that we're not going to Oracle anytime soon and we're
planning to continue pushing the MySQL cluster up from its current 30k
transactions/second data rate toward 100k.  Maybe after that we'll revisit the
question.  If I look at that transaction rate and compare it to rates
supported by the db technologies that I worked with at enterprises 10, 15, 20
years ago:  MySQL blows the doors off all of them even with its current
bugginess.  And at least the bugs have been in areas other than stability; I
get uptimes well past 1 year (we've been running 5.1.x alpha/beta versions
since Sep 2007).

> Fortunately, MySQL is an open source product, so the community can take
> over and continue development in directions that Oracle chooses not to
> pursue.

My hunch is that management at Oracle will do what Bryan Richard said in that
Linux Mag article:  "allow the project to drift rudderless".  It will be a
case of malign neglect:  the more they can sow fear/uncertainty/doubt among
people like me who are facing tough questions from their CTO as to why we
struggle with MySQL's bugginess and/or have to struggle with which of the 3
different forked versions that will come out of all this from advocates not on
Oracle payroll, the more they figure we'll invite that sales team back out
from Burlington to pitch the $2 million enterprise solution.

My belief on all this goes back to something RMS said to me over 25 years ago:
 people shouldn't make money on software.  I don't agree that no one should
ever make money on /any/ software, but I do think that once a reasonable
period has passed after an innovation has been made (such as the 15 or 20
years that a patent provides protection), the invention should pass into the
public domain so the world can benefit and mankind can move on to other things
instead of milking customers of the past.

I think about this every time I open my Verizon bill and ponder why I should
keep writing a $40 check every month for something that ought to be
practically free.  There /are/ other options that most people use but there is
a downside to them that I'd like to avoid.  Isn't it time for basic database
infrastructure to go the way of server operating systems?  20 years ago no one
would have thought there'd be a full-blown enterprise O/S capable of running
circles around every commercial O/S, downloadable and free of charge to anyone
whether they are at IBM or in some village of Rwanda.  *That's* the vision
that RMS had all those years ago and a whole lot of people made that happen.

Oracle does not share that vision, of course.  MySQL is very vulnerable.

-rich