[H-GEN] Windows to *nix

David Jericho david.jericho at aarnet.edu.au
Tue Sep 26 20:36:59 EDT 2006


Robert Brockway wrote:

Hi Rob, I was going to leave this thread alone, but you are a noble warrior, so
to the death we must battle!

Not really, but I'll reply anyway :)

> Apparently I don't.  What features does MS-Exchange[1] have that are
> lacking in the alternatives?

A unified client server interface, and a web interface that offers the
functionality of the client without needing the client. A method by which third
parties can extend and enhance the product with extra plug ins and modules.

The Office suite, Outlook, Sharepoint, and my countless other tools all
integrate seamlessly into it. There is no OSS tool I have ever found that is as
capable.

And it crashes no more frequently than Thunderbird (which is my main MUA).

> It seems to me that the implied argument here is that disk, processing
> power and ram are cheap so it is ok to throw more resources as a problem
> than to make a more efficient algorithm (in your applicatiom, OS, etc).

The thing I keep having to remind people, is stop chasing false economies. The
effort involved with building a machine without X in certain environments, does
not weigh up when compared to the extra cost of purchasing a slightly bigger disk.

And I don't want to see you dare suggest that Linux is more efficient than
Windows, without actual hard real world figures, just because it's OSS.

I'm sure there are Microsoft programmers (in fact, I know a few personally) who
take great offense at the common assumption that they do not take some pride in
the quality and efficiency of their code.

> 1.  It is a false assumption in some situations.  Sometimes the demand
> on the computing resources really is near the limits of the hardware.  A
> less efficient algorithm means reduced performance.
> 
> The budget doesn't always extend to purchasing more hardware.

I love my little naff sayings and philosophies. One of my favourite is "If you
can't afford two, you don't need one."

The glory days of IT are over, I rode that wave and spent most of it drunk as a
skunk. Now, IT is a business requirement, companies without suffer as a result.
I often do resource comparisons between various solutions and so many
administrators make savings through false economy.

How important is a system? Critically important, as in the staff of a company
wouldn't be able to work without it? Critically important on Tuesdays between 11
and 12am because that's when the pays are run, and unimportant otherwise?

> 2.  The argument presumes there is no where else to spend the money.
> This is always false.  There are always other places money can be used.
> 
> In a business this could mean high salaries to retain top employees,
> better non-computer resources in the work place, or even higher profits
> to shareholders.

High salaries for employees are a total waste when the tools they use are
unavailable. It's like having an entire floor of share traders unable to do
their job because the IT systems are down.

> Thus I find it is important to optimise the computer system until such
> time as deminishing returns mean it is unreasonably expensive to do so.

I agree completely.

> If squeezing the last 10% of the speed could take 90% of the effort then
> the first 90% of optimisation would only take 10% of the effort.  The
> first 90% of the optimisation is a good deal.

Once again, I don't disagree, and these are basic modes of operation for any
competent IT professional.

> Anthony Irwin wrote:
>>> You don't have that control in Windows you have to have the graphical
>>> environment you can't choose to strip it down to the level that you
>>> can on a *nix system.
>>
>> Rubbish. I suggest you go investigate the Windows world much closer.
>> Information like that is simply false, and belies a greater
>> misunderstanding of the paradigm.
> 
> Can MS-Windows be run in a production environment without any GUI?  I
> did not think this was possible.  I would be most interested to hear if
> it can be done.  If it cannot be done then Anthony's assertion is correct.

I think you misunderstand. Anthony suggested that the only way to do things was
via the GUI. This is not true.

> Thanks to the GUI running in ring 0 on an IA32 system it does reduce
> system stability.

Microsoft have done an amazing job of actually fixing this. I'm struggling to
actually remember the last time one of my Windows systems, be it personal or
corporate blue screened on me.

I do however have clear memories of when one of my Red Hat boxes crashed.
Infact, I can routinely make Red Hat machines die quite easily. I doubt many
people here on this list have capacity and utilisation figures to match mine,
and I often detune machines for stability with a major impact on performance.

The perception that Windows is an unstable operating system is an old one, based
in the days of Windows 95 and 98, which were unstable.

> Similarly I think the number of services and processes than cannot be
> disabled in an MS-Windows system is much larger than in *nix in general.

This is false. Check a Windows 2003 Server system sometime. The vast majority of
services are disabled, beyond the various core services.

> As always if hard data to the contrary can be presented I will stand
> corrected.

While I would love to spend hours arguing the point, I'll merely suggest you
obtain or get access to a Windows 2003 system. Check it out, it's not the great
evil Linux advocates love to claim.

> Cost is only one aspect of OSS and to consider it without considering
> the other ways in which OSS is beneficial is to misrepresent the true
> costs and benefits, IMHO.

Hence, resources. But many of these other benefits come back to time once again.
Having the source is of no use if I must spend a month figuring out how the
application works.

> It seems to me that much of this post is based on the assumption that
> using OSS will inherently take more time to perform tasks.  This has not
> been my experience but if it has been your experience then you may well
> find commercial software a better alternative.

You have very much misunderstood the intent of my post then. I thought I had
made it clear that it was an analysed approach, with the resource use weighed up.

> Do you still use OSS software much these days David?  Much of this post
> seems to have been about perceived negative aspects of OSS so I suspect
> not.

At this instant, I have 170 Red Hat Enterprise Linux licenses deployed from
Frankfurt, Germany, to Dallas, Texas. I have quite a number more about to
appear. I drive these systems via Cfengine hosted on Linux with revision control
applied via Subversion, on an Apache server, configured using vi inside screen
while logged in using Putty.

Many of those who speak to me daily know I have very little time for blind
advocacy and uninformed opinion. Thankfully most of the OS X fanbois seem to
have crawled back into the dope smoking hippy corner they came from[0]

I'm sure you remember well, my rampant and almost fanatical support of OSS
previously. In 10 years of doing this professionally, I'm now jaded. OSS as
advertised by advocates is a crock. I feel ripped off.

Each time a machine crashes to a kernel bug, no, I don't load up vi and spend my
day being one of the egalitarian technoratti, hacking away on the kernel until
the bug goes away, and then sharing my wonderous discovery to the world.

Infact, I'll be brutally honest. I don't effing care. Maybe it's not the moral
high ground, I can't claim some mighty and noble libertarian ideals, but I do
live very firmly in the real world, where I and my staff want to go home at 4:30
pm on Friday afternoon and know that our services aren't going to disappear on
the weekend.

I prefer not to be blind to the deficiencies of the tools I use because I've
gotten hung up in layer 10, being religion. I am already very well aware of the
advantages.

[0] I consider myself an anti-advocate. Consider the irony.

-- 
David Jericho
Senior System Administrator, AARNet
Phone:     +61 7 3317 9576
Mobile:    +61 4 2302 7185




More information about the General mailing list