Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 25 Apr 2001 12:20:42 -0400
From:      Nathan Vidican <webmaster@wmptl.com>
To:        technet@microsoft.com
Cc:        questions@freebsd.org
Subject:   HotmailMigration
Message-ID:  <3AE6F95A.4D51D58D@wmptl.com>

next in thread | raw e-mail | index | archive | help
	I've got a few questions which were not answered in your article. I
will tell you a little about my business, and what I am currently
working with. I cannot justify moving from my open-sourced system to
Windows 2000, and am curious as to why you would choose HotMail as a
prime example? I do not know of too many companies which could afford
(primarily speaking about financially) the migration of 5000+ servers
from an open-source (cost-free licensed) to Windows 2000. I cannot even
begin to fathom what such a cost would be, based upon my experiences
with Microsoft and it's software licensing charges. I am attempting not
too be biased with this email, as I am dead serious, and would truly
appreciate an honest reply.
	I am currently involved in the startup phase of a new devision to an
existing company. My network is a relatively small network with a
cluster of about 30 servers (and growing), running web, database, and
email services. On all of the intel-based machines I am running the
FreeBSD (various releases from 2.2.8-RELEASE -> 4.3-STABLE) operating
system, on Sun Sparc based machines I am running either Solaris 2.8, or
OpenBSD 2.8 and on all of the Apple Macintosh systems I am running
OpenBSD 2.8 (used primarily for SSL servers). All servers are on a
100mbit switched LAN, (using Cisco Catalyst switches), with Gigabit
(over fiber) connections between several of the key servers (eg: primary
user database and mail servers) in the progress of implementation. We
offer web-hosting, server co-location, and custom web/database design. I
use Oracle, MySQL, and PostGre SQL as well as a custom design which
leverages the UFS filesystem's capabilities for the databases. In
general, I am happy with the design, and really satisfied with the
performance.
	I cam accross your article reading a thread posted to
questions@freebsd.org; and from there I understand that your move from
FreeBSD to Windows 2000 was primarily motivated by political reasons. I
can see how it would be a huge embarassment to be running one of the
most successful sites on the internet from a platform which you cannot
market and/or sell yourselves. Still, from an unbiased perspective, I'd
ask you to answer a few questions regarding this migration.
	First and foremost, I can see how this migration wouldn't have incurred
the licensing charges as it is a part of Microsoft; but how much would
the total cost for the licensing on all of the HotMail webservers cost
the average business (which is not owned nor operated by Microsoft)?
Judging by what it would have cost me to run Win2000 Server for a measly
25 users on our local office LAN I can't even contemplate the licensing
cost for 5000+ servers serving millions of users. I do understand that
one may be able to get away without purchasing so many copies of the
O/S, but I cannot see how you would legally allow an outside company to
run that many servers without individually licensing them. To be honest,
I do not know much about Microsoft's licensing policies... perhaps you
do offer some sort of bulk, flat-rate, or site-wide licensing fee
schedule for large server farms of this nature, do you? If so, what
would the MSRP be for say 5000 servers each serving 2000 users a piece?
	Secondly, what about the upgrade path? What kind of options would you
be able to offer someone with a 5000 unit server farm when you're next
'Service Pack' comes out? Would that company then have to 'upgrade' all
machines to the new service pack in order to avoid some threatening
back-door entrance to the system that you 'weren't aware of'? What about
say the next generation of the O/S? Would the company then have to
re-purchase licensing for each server to be upgraded? What would this
cost? I mean just the software cost; not including the IT staffing
requirements to actually procure such a transition.
	Thirdly, to what level of customization could you possibly offer to
such a company? With FreeBSD, (or any other Open-Sourced system), the
company would have the extreme flexability in that they could simply
change or alter components of the O/S which they simply do not require.
For example, the GUI; similar to Novell Netware 5, wherein the GUI
doesn't have to be using system resources because it doesn't always have
to be loaded... or ever loaded to begin with for that matter. What if
the company were involved in some obtuse practise, (say VOIP), wherein
they were required to service thousands or even millions of clients over
a protocol which your O/S doesn't directly support? Would they then have
to fork-out more money for OS 'add-ons', and/or 'upgrades'? Would they
then incur more licensing fees? Would they be forced to go to a third
party to provide them with a solution that would work on your platform?
Would you then refuse to support it because it was designed a third
party; worse yet, would you blame problems their encountering on
imcompatabilities or state that you 'do not support that'?
	Which brings us to another issue, support. What would you charge for a
support contract to be able to fully provide technical support for a
company of this magnitude? I know of several firms which support FreeBSD
for such massive endeavors, but I would assume that something this large
would have to be supported by Microsoft directly. Primarily so because
no-one else (should) have more knowledge of the O/S and/or how it
functions; and therefor be knowledgable enough to support it. What would
a typical technical support contract for a company with 5000 servers
running Windows 2000 cost?
	Lastly, what about performance. You stated in your article that FreeBSD
does not perform well in a given scenario. From what I understand, that
given scenario really doesn't apply to Apache. Therefor you compared the
performance of two completely different things, of course stating that
yours was better. My question is, that if yours is so much better then
why are some of the largest sites on the internet still using FreeBSD
and Apache? These are the few companies which this migration could
potentially apply to, for example Yahoo. Companies which do service as
many if not more clients than HotMail, yet I don't see them even
attempting to justify such a migration. Why do you suppose they aren't?
Have you approached companies like Yahoo? What was the typical response
given to you by them?
	To sum up things a bit, I would like to know what it would have cost an
outside company to make such a transition from and open-sourced system
to Windows 2000. This would of course not include their IT staffing,
programatic changes, and downtime incurred to do so. I would like to
know why, (other than for the obvious political reasons), you chose to
make such a migration public news. I would like to better understand how
you could possible intend to market such a large migration for any other
company outside of the Microsoft Corporation. If you could reasonably
answer any of the above, in a manner as unbiased as possible I would
truly like to hear from you. Please note that I have also carbon copied
this message to the FreeBSD support list, in an effort to understand the
FreeBSD community's viewpoint on this as well. I would encourage you to
also carbon copy your response to this list, as I am sure I am not the
only one with questions of this nature.

Nathan Vidican
Nathan@Vidican.com
http://Nathan.Vidican.com/

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




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3AE6F95A.4D51D58D>