• 1

To the contrary

(Anonymous)
SUSE had determined to drop SELinux long before AppArmor came along, because it had proved to be unusable. Even RH users seem to think so, as RH reported at the 2005 SELinux summit that the #1 question was "how do I turn SELinux off?"

SELinux has been available to the open source community for many years, and is now a standard part of RH's install. Yet the vast majority of people who look at it reject it and choose nudity instead :) In the usual open source world, that rate of rejection usually says "nice try, how about something else?"

Contrary to the claims above, I think AppArmor is actually *more* flexible than SELinux. Because of its much greater ease of use, a mere mortal sys admin can adjust AppArmor policy on the fly, without even having to re-start the contained process. Changing SELinux policy, in practice, requires someone with substantial SELinux expertise. Unless you have someone with those skills on staff, you need to call a consultant just for a configuration change.

Your fears about application developers having to choose between SELinux and AppArmor are unfounded; unlike SELinux, AppArmor does not require the developer to support AppArmor. Anyone can make an effective profile for an application with the AppArmor tools.

The reason for the split is that this has nothing to do with GUIs. For that matter, I'll concede that SELinux actually has prettier GUIs, but AppArmor is still an order of magnitude easier to use. The reasons are deeply founded in the security model, and it would not be possible to move SELinux to this point without completely ripping it apart amd making be ... well, AppArmor. SELinux and AppArmor are already sharing the code that we agree upon: LSM http://lsm.immunix.org/ which was developed jointly by Immunix and SELinux, as well as numerous other open source contributors.

I am painfully aware of the UNIX fragmentation wars. Open source licensing is the solution, which is why AppArmor was recently released under the GPL http://en.opensuse.org/Apparmor Competing projects in the open source space allows users to make a choice. With the very low adoption rate of SELinux, I think users are crying out for another choice.

Crispin
---
Crispin Cowan, Ph.D. http://crispincowan.com/~crispin/
Director of Software Engineering, Novell http://novell.com
Olympic Games: The Bi-Annual Festival of Corruption

Re: To the contrary

(Anonymous)
This is the second time I hear something like this from Novell:

"unlike SELinux, AppArmor does not require the developer to support AppArmor"

Novell Users FAQ (http://en.opensuse.org/Users_FAQ) says:

"Applications don't have to be modified at all to be protected by AppArmor. To get the full power of SELinux, applications must be recompiled and linked against SELinux libraries."

I'm not sure what you mean by "to get the full power", but above statements seem to be a blatant lies. Could you please elaborate what modifications are necessary for the appliciation to be protected by SELinux ? Propaganda is a subtle art and exaggerating may be dangerous.

Re: To the contrary

(Anonymous)
It's pretty much a lie, which is why he hasn't responded. :(

OTOH, we don't have to worry about apparmor going far if they have to resort to fibbing to make it look advantageous.

Re: To the contrary

(Anonymous)
I'm just curious if any of the responders have actually used and implemented both products. I've worked with AppArmor, SELinux, and Argus Pitbull under Linux. They all offer significantly improved security. The difference is in how long it takes to secure something. If you have weeks to spend securing something with SELinux or Pitbull, great. I'd rather spend an hour with AppArmor and be done with it. I don't work for Novell but I've known the Immunix guys for a long time. Check out LSM (http://lsm.immunix.org) if you think they haven't been involved in everything for a long long time.

-Dan Elder
Linux Practice Manager
Novacoast, Inc.

Re: To the contrary

(Anonymous)
Excellent, but does your opinion have anything to do with my original post ? I didn't say that AppArmor or Immunix are bad or incompetent, just that Novell seems to spread deliberate misinformation about SELinux, that's all.

I can believe AppArmor is easier to use (I use SELinux and only read AppArmor manual) but I also suspect (perhaps I need to learn more) that it's also less flexible and possibly less secure. MAC shouldn't be considered a toy, for which pretty GUI (and no need to training) is the key factor. To be perfectly honest, I admit that SELinux documentation should be definitely improved.

Re: To the contrary

(Anonymous)
In a fully deployed scenario SELinux certainly does have the ability to provide greater security than AppArmor as it can address a wider range of threat models. Unfortunately, making MAC both secure and practical (i.e., deployable for most organizations) isn't the easiest of task and AppArmor does focus on usability instead of covering every possible threat vector. I didn't mean to discredit SELinux, just to point out that for the vast majority of users out there for whom SELinux is too complicated to deploy that AppArmor is avilable and very easy (in my mind anyway) to setup.
Cheers,
Dan

Re: To the contrary

(Anonymous)
Actually, to be fair to Crispin...I believe he's been traveling quite a bit over the last month...probably more so than he's used to. Know I've seen him at a few locations across the US this month and I've been traveling a bit...he might have missed this...don't think he'd avoid confrontation to be honest :-)

But please feel free to post the question to him directly or to the public forum set up for AppArmor (which I believe he may watch a bit more than this blog...no offense) and you'll probably get the answers you need.

If you think they're wrong, by all means, ask them to show their proof or stop saying it. But honestly, most of the folks I've known at Novell don't blatantly lie...it's really not in their nature...most are actually hard working individuals just like at other organization trying to make a difference, e.g. RedHat, Apple, etc.

Re: To the contrary

SELinux has been available to the open source community for many years, and is now a standard part of RH's install. Yet the vast majority of people who look at it reject it and choose nudity instead :) In the usual open source world, that rate of rejection usually says "nice try, how about something else?"

Have you considered the possibility that these users have no need for either SELinux or AppArmor? I'm one of the people who turns SELinux off. It's not because it's flawed. It's because I got an EACCES one day, did not understand why from "ls" output, and then much later saw in /var/log/secure something about SELinux denying the call. It was easier to turn it off than to understand an entire layer of security that I didn't need on a minimum-security installation. The standard Unix model was sufficient for what I was doing, and I already understood it. AppArmor would have suffered the same fate.

This is probably the common case. These people are not saying "I extensively evaluated SELinux, was dissatisfied, and thus wanted to turn it off". They're saying "I want to use the simple system I already understand". If people don't have complex needs, you won't get them to use a complex security system. That doesn't mean the complex system is wrong. There may be a subset of people who need it to be just how it is, and they'll be pissed if you try to get them to use anything else.

Re: To the contrary

delete

  • 1
?

Log in