Re: Postgres refusing to use >1 core

On Wed, May 11, 2011 at 1:53 PM, <gnuoytr(at)rcn(dot)com> wrote:

> So, the $64 question:  how did you find an engagement where, to bend Shakespeare, "first thing we do, is kill all the coders" isn't required?  This RBAR mentality, abetted by xml/NoSql/xBase, is utterly pervasive.  They absolutely refuse to learn anything different from the COBOL/VSAM messes of their grandfathers; well modulo syntax, of course.  The mere suggestion, in my experience, that doing things faster with fewer lines of code/statements in the engine is met with overt hostility.

It really depends. For a lot of development scaling to large numbers
of users is never needed, and it's often more economical to develop
quickly with a less efficient database layer. In my last job all our
main development was against a large transactional / relational db.
But some quick and dirty internal development used some very
inefficient MVC methods but it only had to handle 45 users at a time,
max, and that was 45 users who accessed the system a few minutes at a
time.

I've seen EVA systems that people tried to scale that were handling
thousands of queries a second that when converted to real relational
dbs needed dozens of queries a second to run, required a fraction of
db horsepower, and could scale to the same number of users with only
1/10th to 1/100th the database underneath it. In those instances, you
only have to show the much higher efficiency to the people who pay for
the database servers.