Very impressed with MS Intelligent Message Filter for Exchange 2003

  • Thread starter Deleted member 19858
  • Start date
D

Deleted member 19858

Guest
About a week ago there was a thread about the MS Intelligent Message Filter for Exchange 2003. No one really had much to say so I went ahead and installed it since I have a temp server up now anyways. So I wanted to just give my input on it in hopes someone would find them useful.

I installed the filter which was a simple install and took about 5 minutes. After reading through the deployment guide from MS and understanding how it worked I configured the settings and had it running in about another 5 minutes. I set the level to 6 and had it archive the messages until I am comfortable to let it delete them.

I have been using the filter now since the 15th and it has caught almost 1,800 spam messages. I just finished looking through the archive DIR with Outlook Express and from all of the mail I looked at none of them were legitimate. I have a fairly small Exchange system with only about 20 users but I think even on a larger level this would still be a great spam solution.

I like a lot of MS products but with something like this I really was expecting it to do a horrible job and grab a bunch of legitimate email. Also with it being free I expected even less from it since MS doesn't like to give much away. So overall I would highly recommend this to anyone running Exchange that wants a free solution that is simple to install and administrate.
 
SnowPunk98 said:
I like a lot of MS products but with something like this I really was expecting it to do a horrible job and grab a bunch of legitimate email. Also with it being free I expected even less from it since MS doesn't like to give much away. So overall I would highly recommend this to anyone running Exchange that wants a free solution that is simple to install and administrate.
Click to expand...

The downsides to the "free" filter are that the rules are relatively static. Hence they may work for one organization or person, and not another. Since the rules are static, over time those rules will become outdated thus allowing spam to pass by without some kind of update from Microsoft to which they have not stated *if* or even *when* they plan on releasing any updates for it.

IMHO I would only use it as a "temporary" solution while looking for alternative products like GFI Mailessentials.
 
Or a front end with spamassassin on it. That's the way to go, I think. Custom rules, a large community writing rules, simple to install the new rules. Throw in autowhite-listing and the bayes autolearn, and you've got yourself a pretty spiffy little email server.
 
I am running Exchange 2003 with SP1 installed

I have also used GFI MailEssentials and found it to be pretty good program. When I had the trial version running on my server I was not able to get the Bayesian filter to work at all so I never took advantage of that. Possibly having that as a contributing factor I did get a good number of false positives. Also GFI is gonna run you almost $400 for the size of exchange server I have and I don't have that kinda money to spend.

Yes eventually since the rules are static there will probably be workarounds but they will probably also affect other spam filers. Hopefully MS will stay up to date with this and offer more free updates to it.

As far as installing SpamAssassin as a front-end this would be great except for the fact that it would require another server running Linux, this means hardware and another static IP not to mention space. I am trying to look at this in terms of a cost benefit analysis and wanted to offer my thoughts.
 
SnowPunk98,

That's the trick to Bayesian filtering, *you* have to teach it what is good email vs bad email, it takes awhile at first, but once the filter is in place not only does it do an excellent job, but it *continually* learns over time. With Bayesian filtering you will get some false positives, but they can easily be added back to GFI's good mail list.

Since MS IMF is static, if it does flag an email as spam, you don't have anyway of telling IMF that its good mail other than putting it into a white list which is not always a good idea. Bayesian is not subject to stale rules like MS IMF.

$400 is very respectable in terms of GFI's product offering, I can't find anything else that compares in price,performance, and features. While I will agree that IMF is a decent free offering, it should not be considered a permanent solution unless MS somehow adds more capabilities to offset the relative "static" nature of its analysis.

Just my humble opinion. :p
 
I totally agree with you on what you said. For me though running a very small exchange server I cant afford $400 for GFI, I would if I could. Bayesian is the way to go no doubt about it.

With GFI is it possible to setup rules for individual mailboxes or is it just global?
 
SnowPunk98 said:
I totally agree with you on what you said. For me though running a very small exchange server I cant afford $400 for GFI, I would if I could. Bayesian is the way to go no doubt about it.

With GFI is it possible to setup rules for individual mailboxes or is it just global?
Click to expand...

There are certain things you can do on a per mailbox basis, give me an example and I will see if it is possible.
 
1. Say there is a user who does not want his mailbox scanned for spam because he wants to do it client side. Can you exclude him from the global scanning?

2. Would it be possible to set certain accounts to delete spam while others have their spam forwarded to a junk folder in their mailbox or forwarded to another email address?
 
SnowPunk98 said:
1. Say there is a user who does not want his mailbox scanned for spam because he wants to do it client side. Can you exclude him from the global scanning?

2. Would it be possible to set certain accounts to delete spam while others have their spam forwarded to a junk folder in their mailbox or forwarded to another email address?
Click to expand...

1. No
2. No

...... AFAIK there isn't a spam filter that is capable of being that granular.... not without being *very* expensive
 
You must log in or register to reply here.
Back
Top