Re: [spam] Re: [mail] Re: Native Win32 sources - Mailing list pgsql-hackers

Hannu,

Using a Win32 platform will allow them to perform relative metrics. I'm not
looking for a statement saying things are x per cent faster than production,
I'm looking for reproducable evidence that an improvement offers y per cent
faster performance than another configuration on the same platform.

The QA environment is designed to do final testing and compiling definitive
metrics against production systems, what I'm looking for is an easy method
of allowing developers to see the relative change on performance for a given
change on the code base.

I'm fully aware that they'll still have to use the config files of
PostgreSQL on a Win32 port, but the ability to edit the config files, modify
sql dumps to load data into new schema, transfer files between themselves,
and perform day to day tasks such as reading Email and MS-Word formatted
documents sent to us using tools that they are currently familiar with is a
big plus for me.

The bottom line is I can spend money training my developers on Linux and
push project deadlines back until they become familiar with it, or I can
obtain a free database on their native platform and reduce the number of
machines needed per developer as well as making the current DB machines
available as the main machine for new staff. The latter makes the most sense
in the profit based business environment which I'm in.

Al.

----- Original Message -----
From: "Hannu Krosing" <hannu@tm.ee>
To: "Al Sutton" <al@alsutton.com>
Cc: "scott.marlowe" <scott.marlowe@ihs.com>; "bpalmer"
<bpalmer@crimelabs.net>; <pgsql-hackers@postgresql.org>
Sent: Wednesday, November 27, 2002 10:54 AM
Subject: [spam] Re: [mail] Re: [HACKERS] Native Win32 sources


> On Wed, 2002-11-27 at 08:21, Al Sutton wrote:
> > The problem I have with VMWare is that for the cost of a licence plus
the
> > additional hardware on the box running it (CPU power, RAM, etc.) I can
buy a
> > second cheap machine, using VMWare doesn't appear to save me my biggest
> > overheads of training staff on Unix and cost of equipment (software and
> > hardware). I've been looking at Bochs, but 1.4.1 wasn't stable enough to
> > install RedHat, PostgreSQL, etc. reliably.
> >
> > The database in question holds order information for over 2000 other
> > companies, and is growing daily. There is also a requirement to keep the
> > data indefinatley.
> >
> > The developers are developing two things;
> >
> > 1- Providing an interface for the companies employees to update customer
> > information and answer customer queries.
> >
> > 2- Providing an area for merchants to log into that allows them to
generate
> > some standardised reports over the order data, change passwords, setup
> > repeated payment system, etc.
> >
> > Developing these solutions does include the possibilities of modify the
> > database schema, the configuration of the database, and the datatypes
used
> > to represent the data (e.g. representing encyrpted data as a Base64
string
> > or blob), and therefore the developers may need to make fundamental
changes
> > to the database and perform metrics on how they have affected
performance.
>
> If you need metrics and the production runs on some kind of unix, you
> should definitely do the measuring on unix as well. A developers machine
> with different os and other db tuning parameters may give you _very_
> different results from the real deployment system.
>
> Also, porting postgres to win32 wont magically make it into MS Access -
> most DB management tasks will be exactly the same. If your developer are
> afraid of command line, give them some graphical or web tool for
> managing the db.
>
> If they dont want to manage linux, then just set it up once and don't
> give them the root pwd ;)
>
> --------------
> Hannu
>
>
>
>