1. Skip to navigation
  2. Skip to content
  3. Skip to sidebar
Source link: http://archive.mises.org/10934/database-tweaks-toward-the-light/

Database Tweaks Toward the Light

October 29, 2009 by

Yesterday late, you might have noticed the inability to comment on the blog. Maybe you noticed the site go down and up ever so briefly. There was order emerging amidst the madness.

The wonder of it all is that with a good development team, every seeming problem is a foreshadowing of a more perfect world to come. The bigger the calamity, the greater the leap toward progress.

In this case, the development was pushed forward by a little database stick or lock that has vexed us for up to a year. Only two weeks ago, the problem became persistent and somewhat predictable. About every ten minutes, there was an 8 second delay in the response time of the database.

Now, if you know something about this subject, you know that this area is an extreme specialization, one that requires a rocket-science level of knowledge on the particulars. The all-round amazing David Veksler saw the need for such a mind, and solicited help on this blog.

So many great people responded, but we let Jordan Bullock take a crack at it. I loved receiving his constant technical updates even when I couldn’t understand a word in them. It’s kind of like reading magazines about sports cars you can’t own: we experience a vicarious thrill just from looking.

So on it went for two weeks: new tools, new tweaks, new arrangements and configurations, new tests, new efficiencies. In the course of doing this, David was also busy on the hardware front. He wanted the newest supermega database server, one which, it turned out, wasn’t even scheduled for release until two days ago.

It became a matter of timing: perfect the structure in time for the move. Yesterday, it was all set. Jordan finished his work, and Isaiah Houston, our infrastructure specialist, turned the switch, and, in minutes, a new world was revealed unto us.

Today we wake to the dawn. It is a wonderful thing to behold. It seems like the stick or lock in the database is no more. In the end, it was not one problem or one process or one file, but a series of interactions that needed slight improvement up and down the line, together with some new hardware.

Was it the software improvements or the hardware that made the difference? David describe it this way. A car (software) with engine problems is going to have an easier time of it driving on a highway (hardware) that isn’t clogged with traffic. Meanwhile, a faster car is going to navigate even difficult highways better than a problem car. If you improve both, it is the best of all worlds. It seems like we achieved improvements in both.

The key to going from a persistent problem to a new and more perfect state of being was a process, and continuum of development, a long series of small changes: changes on the micro level and not the macro level, changes at the margin, not the aggregate; not central planning but a series of tiny steps taken with deliberation and discipline.

Now, if someone in Washington would consult David, Isaiah, and Jordan on how to move the macroeconomy from darkness back into the light.

Reblog this post [with Zemanta]

{ 6 comments }

Conza88 October 29, 2009 at 9:54 am

So that is why everything is faster! :D

Matt October 29, 2009 at 10:15 am

I have no idea what this means, but I am glad Mises.org is up and running better than ever.

DixieFlatline October 29, 2009 at 11:20 am

Three cheers for the heroes of the database!

filc October 29, 2009 at 4:01 pm

Thinks doo feel faster. I thought it was just me!

theBuckWheat October 29, 2009 at 10:27 pm

Congrats. DB freezes are very hard to nail down and fix. Now, while you are basking in the glow of your new server, could you make at least some of the publications and media available for private mirroring?

Haas October 30, 2009 at 5:55 am

wait testing testing

Comments on this entry are closed.

Previous post:

Next post: