MS Exchange server on FreeBSD?

[Bart Silverstrim]

On Mar 20, 2005, at 1:18 PM, Anthony Atkielski wrote:

> Duo writes:
>
>> And you failed to answer his question. Why not stop trying to avoid 
>> it by
>> answering it.
>
> I did answer it.  I asked for a product that provides ALL the features
> of Exchange.  And he surely knows what all of the features of Exchange
> are, otherwise he could not say with confidence that other UNIX 
> products
> provide them.

Why when this type of question comes up is it so hard for someone just 
name those must-have features that are so necessary?  It always sounds 
like a cop-out...

A: This solution is best.
B: I disagree.  I can do it well with this solution.
A: Bah.  That solution doesn't have all the features and convenience of 
my solution.
B: Oh yeah?  Bet my solution can do it.  What features are missing?
A: Look, you're intent on trashing my solution, so you must be familiar 
with it.  I want to see a solution on your platform that has the exact 
featureset of my solution.

I mean, even if he named a perfectly useable solution, what would you 
do then? Counter that it doesn't include a helpful Office Assistant and 
full integration with an office suite product?

"I'm looking for a car that looks, acts, feels, and drives just like a 
Ford Focus.  It's the best in the world.  I challenge anyone out there 
to come up with a car like it..."  If you have an exact product in mind 
and aren't going to be happy until every last menu and interface glitch 
and memory footprint-hogging feature is matched point for point...buy 
that product and let it be.  We run an Exchange server, and out of a 
thousand users, we have maybe five people that actually use the 
calendar function.  Everyone else uses it purely for email.  Waste!  
The only things we use out of it are IMAP/Exchange interface fore mail 
retrieval, SMTP for sending, five people use the calendar, address book 
that sometimes people actually try using without subverting with local 
personal address books that they're duplicating addresses (often with 
typos) that already exist in the global directory, integration with AD 
(one of the only useful things I've found in using it on a 99% Windows 
network...create their account, they already have email ready), and the 
web interface.  While it may work like a charm for you, I find the 
header mangling to be a major PITA when troubleshooting things.  I love 
it more when I'm using my favorite email client (Mail.app) to read mail 
and retrieve bounces, and most of the info regarding the email (or 
attachment) is mangled or stripped so I have to use Outlook from an RDP 
session to "re-send" the message just so I can see the actual content.  
That's a feature?  I find it to be a PITA.    And what should we do 
when people have preferences for different clients for tasks?  With 
Exchange/Outlook, you get an all or nothing.  Pretty much anything else 
is crippled or somewhat usable unless you're running Windows with 
Outlook on it.

Want an ultimate solution?  Create a server platform that is modular 
and is interfaced by API, so your client can be running on any platform 
or interface and still get the same information.  Don't use a server 
"system" that is so tied to a particular product that everything else 
is crippled or supported only part way.  Maybe "cobbling" together a 
solution may be better for some institutions.  I know that for me (and 
many of our client systems) the bandwidth requirements and memory 
restrictions on the desktop make running that bloated memory hog 
Outlook far more a pain than a boon while trying to get other things 
done.

>> As for looking for non microsoft solutions, yes. There is a point to
>> that. It's called voting with your pocketbook, and its a valid course
>> of action in a capitalist society. Choosing to go outside a monopoly
>> is a right.
>
> So you never buy Intel microprocessors, and you never buy anything with
> a zipper?  (Remember, YKK has a virtual world monopoly on zippers.)

Intel isn't a monopoly, since AMD seems to have a large stake in the 
market running the same stuff Intel chips run.

As for zippers, I'm pretty sure the interface is well documented, and 
if you want to come up with a better design to market, you're probably 
free to.  If they used their position in the market to lock you out, 
like, say, threatening their buyers with retribution if they were to 
purchase your product, then that is abusing the monopoly power.

Monopolies aren't necessarily bad, it's when they abuse that position 
that it is a problem.  If nobody wants to challenge the current 
monopoly holder as having the best pooper scooper on the market, it 
makes little sense to cry monopolist over it.

>> And yes, looking for non MS solutions, for the sake of it, is a valid
>> choice.
>
> Not for many corporate managers.  They don't care whether it's 
> Microsoft
> or not, as long as it's the best tool for the job.

And when they're too ignorant to do what their IT people tell them to 
choose instead of what "other people are doing"?  Best tool for the job 
my arse...Name one IT admin who isn't getting "political pressure" to 
choose a particular solution at some point in their professional 
career.

Pick up the latest Linux Format.  It has a fun article on the Navy 
choosing Win2k to run their latest ships.  It has some quote in it from 
the American Navy having problems and saying that they were under 
political pressure to choose it (Windows NT 4.0) despite some 
reliability issues.  Best tool for the job?

>  People don't usually
> reach the upper levels of management in large corporations by indulging
> emotional attachments to one vendor or another.

I've heard more than one that enjoys the benefits of choosing 
particular vendors while overlooking minor shortcomings.  Dinner here, 
free toys there.  Sometimes it's a vendor willing to dedicate a 
salesman that strokes the ego the right way.  And many upper-management 
types don't dedicate as much time learning about the trenches anymore.  
They're too busy managing.

Worse yet, solution pits.  I had someone tell me that a mail server 
that went wonky couldn't be replaced with a free solution.  Why?  
Because they already invested so much in the current (and at the time 
broken) solution.  Ten grand in licensing per year at the time just to 
run it.  Plus the cost of getting it fixed after it decided to go 
south.  Hmm...free, versus thousands a year.  For just doing 
POP/IMAP/SMTP.  Best tool for the job?...

>> If for instance, I go with a product of MS, as opposed to a smaller
>> OSS project, the OSS Project typically *cares* about the feedback I
>> give it. It cares about the features I want and need.
>
> So does Microsoft.  That's how it stays on top.

So, the overwhelming market share and people's reluctance to "have to 
relearn" things doesn't play any role in this, right?

> It's all a bit amusing, since I remember when Microsoft was the 
> underdog
> and the Great Satan was IBM or DEC.  The names change, but the game
> remains the same, and the flying accusations are just as baseless today
> as they were back then.

Google for "Why I Hate Microsoft", a personal rant that is a very good 
read.  Highly recommended.

> It's a pity that no discussion of software can be carried out these 
> days
> without degenerating into religious jihads against Microsoft.

Personally, I give credit where credit is due, in my opinion.  There 
are things Microsoft didn't do poorly.  But there are many things that 
can and should be redone.

>> I need a credit card before MS will talk to me.
>
> You need a stroke of good luck before someone working on open source
> will talk to you.

Depends on the project.  I had an amazing response to problems with 
PortManager, from the author himself.  And on this list, there's 
apparently an ongoing flamethrower demonstration with the head honcho 
himself of OpenBSD.

>> The Exchange solution might be best for a gold partner with M$, but
>> overall, a very poor solution, which locks you into a feature set, and
>> a company that has shown little concern for its base of customers.
>
> The success of the product would seem to belie your claim.  A lot of
> organizations and users really like Exchange.

And others, like us, are using it pretty much because we are told we 
must, and try to make the best of it.

>> This is a problem for things such as virus scanning, and tight
>> integration with an AD Environment, which is getting more and more
>> replication based. In fact, some types of virus scanning can introduce
>> data corruption of the store, which could lead to other issues.
>
> Step number one in any Exchange database failure is to turn off and
> deinstall all the antivirus junk running against it.
>
> I'd tend to prefer to put antivirus stuff on the client, not on the
> server.

Nice, but expensive at the desktop.  If you can stop the incoming 
intruders at the front gates, it's generally less hassle to do that.  
Besides, with the constant communications back to the Exchange server, 
why not put it on the Exchange server?

But with the stability problems these can introduce (and the already 
bloated requirements to run it), I can see your point.  Maybe that's 
why we moved incoming mail to a pre-screening mail server (FreeBSD) 
that scans initially for spam and viruses then hands it off to the 
internal mail server.  That reason, and the security problems that seem 
to crop up with running Exchange on Windows kept us from wanted it to 
run unshielded.

> Some users may not want their e-mail scanned for viruses.
> Power users, in particular, may not want any virus protections at all,
> since they know not to click on attachments and antivirus software all
> too often hashes the very system it's supposed to protect.

The only reason I didn't want mine scanned was to get samples to test 
with.  Other than that, it was annoying as hell to keep deleting the 
same fake Windows-patches messages and cruft.

>> What's more, the virus scanners that do run against Exchange's DB,
>> also cost money, and typically require some more hardware. And
>> overhead. So now I am running exchange, and a bevy of other stuff to
>> prop it up.
>
> You don't have to run virus scanners.

When 99.9 percent of your users aren't computer literati, this is 
unacceptable...you need to protect them from themselves as much as 
possible.  Statements like that one you made will cost you credibility.