Skip site navigation (1)Skip section navigation (2)
Date:      Tue, 10 Apr 2001 16:12:16 -0700
From:      Alfred Perlstein <bright@wintelcom.net>
To:        Eric Lee Green <eric@estinc.com>
Cc:        r.hyunseog@ieee.org, hackers@FreeBSD.ORG, advocasy@FreeBSD.ORG
Subject:   Re: Interesting article.
Message-ID:  <20010410161216.G15938@fw.wintelcom.net>
In-Reply-To: <Pine.LNX.4.21.0104101535000.980-100000@england.inhouse>; from eric@estinc.com on Tue, Apr 10, 2001 at 03:39:06PM -0700
References:  <3AD2FF23.239CD9DF@moonworld.org> <Pine.LNX.4.21.0104101535000.980-100000@england.inhouse>

next in thread | previous in thread | raw e-mail | index | archive | help
* Eric Lee Green <eric@estinc.com> [010410 15:38] wrote:
> On Tue, 10 Apr 2001, HyunSeog Ryu wrote:
> > . Migrating Microsoft® Hotmail® from FreeBSD to Microsoft Windows® 2000
> > Technical Case Study  
> >    http://www.microsoft.com/technet/migration/hotmail/default.asp 
> 
> They forgot to document reason #0 for migration to Windows 2000: Because
> running FreeBSD made people laugh at their salesmen whenever their
> salesmen suggested migrating from Unix to Windows 2000. "If Windows 2000
> is so good compared to Unix, why is Hotmail still running FreeBSD?".

The Migration took them over 2 years!

(from http://www.microsoft.com/technet/migration/hotmail/default.asp):
    Microsoft Hotmail web-based e-mail service is one of the most
    widely adopted free e-mail services available on the Internet.
    Acquired by Microsoft in 1997, with an installed base of 9
    million users,...

The Migration wasn't completed until after USENIX 2000.

If they had any intention of being fair, they would have also
mentioned that:

1) The migration took over a year (most likely two years) because
they needed to constantly retune Windows to perform as well as
FreeBSD even though they state:

       (from http://www.microsoft.com/technet/migration/hotmail/hotplan.asp)
       "
       The reasons for converting to Windows 2000 were:

       1.Performance (and therefore cost). The cgi "one-process-per-socket"
       model, under FreeBSD, is very inefficient. Per machine
       throughput can be dramatically improved by moving to a
       multi-threaded application model. This results in one of
       the following conditions:

         o) A fewer number of servers required to support the site
         o) Support for a greater number of users by using the servers
            already deployed at the site
       "

   I'm sure if they added the cost of the redesign and hacks they
   had to do (cost of an entire software/ops team for about a year)
   it would not be cost effective.

   Nevermind that fact that it would have been impossible without
   source licenses from Microsoft for a bunch of tools.  Something
   nearly impossible to attain without great cost if you're not
   actually Microsoft. :)

2) There was nothing forcing them to use CGI under FreeBSD, they
could have migrated to Fast-CGI or a threading model under FreeBSD.
They choose to go nuts and over-spend on this silly migration which
was nothing more than a PR move.

3) Effectively once Mircosoft took over Hotmail, all developemnt
efforts on the FreeBSD side were halted, this obviously made the
point mentioned in #2 impossible.  In fact there were some efforts
to make FreeBSD outperform the new NT hacks, but they were not
allowed to be deployed.

Basically a _MASSIVE_ effort was undertaken at great expense to
replace FreeBSD.  Something that would most likely not be possible
by your average web site without much handholding from Microsoft
as well as over two years of work to undertake such a migration.

With Microsoft holding your hand, they've still got a spare hand
for reaching into your wallet. :)

Actually, if you read the footnote F2:
  "
  F2 - There obviously are multi-tasking/multi-process solutions that
  Hotmail could have leveraged under FreeBSD.  However, they would
  require making application modifications and rework to implement.
  So, this was an optimum opportunity to examine other options and
  platforms.
  "

Trully this was a cost effective and justfied reason to switch
out the underlying OS instead of doing a couple of hacks to the
already deployed applications.

Of course I'm being sarcastic.

-- 
-Alfred Perlstein - [bright@wintelcom.net|alfred@freebsd.org]
Instead of asking why a piece of software is using "1970s technology,"
start asking why software is ignoring 30 years of accumulated wisdom.

To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-hackers" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20010410161216.G15938>