[besson3c]

1) Provide some sort of incentive to run OS X Server at all. In big enterprise environments, the just-add-water design approach just doesn't work. Each enterprise has unique approaches to account creation/maintenance, authentication, and in many cases this involves sharing data with other groups. OS X Server is designed primarily to make manual modifications to configurations easy, but these GUIs have no value to organizations needing flexible, automated workflows. In high performance computing environments, you wouldn't want the overhead of running any GUI at all.

2) Taking Apple's GUIs out of the picture, Apple does not offer any compelling open source package management. We can dismiss this by saying that we have Macports and Fink, but anybody that has used FreeBSD, Ubuntu, Gentoo, Redhat, or any number of other systems extensively know that both Macports and Fink are very weak by comparison.

3) As stated eloquently here virtualization is all the rage. OS X server is not supported as a guest OS, and it makes a crummy host OS with the products currently available too.

4) Apple would need to revamp their whole software support infrastructure to assist with supporting leaving the comfortable little Apple bubble and getting into the nitty gritty of low level software issues in a multi-platform Unix centric environment

5) Apple needs to either innovate in this area, or else revamp their whole open source philosophy and commitment to work better within the OSS community

6) Apple needs to stop being schizophrenic in how it designs its own software. Some pieces of OS X support Unix conventions nicely, but others don't. For instance, there is no way to start/stop Apple's VNC solution or application firewall remotely without a GUI. As stated, in many environments the admins are not going to want to run the GUI at all, but are certainly not going to be interested in being dependent upon VNC for remote administration - especially when dealing with clusters of machines. Also, if Apple insists upon developing proprietary solutions over open standards, they need to provide a compelling reason for this. They did the right thing by opening up launchd (and I'm assuming it is documented too), but not so with Screen Sharing and the Application Firewall

7) Software Update is brain-dead and needs to be re-thought for this area. An admin running a web or database server does not care about Quicktime updates, and definitely doesn't want to have to plan to restart the server after these updates have been applied. Chances are, they don't care about iPod software updates either. One fix here is to provide the option to not receive notifications for these things, but for point updates Apple may want to look at ways to prevent having to restart the machine unless the kernel has been modified, or else break up the point updates into smaller pieces allowing admins to only install the pieces of these that they are interested in.

[mduell]

Originally Posted by besson3c 

2) Taking Apple's GUIs out of the picture, Apple does not offer any compelling open source package management. We can dismiss this by saying that we have Macports and Fink, but anybody that has used FreeBSD, Ubuntu, Gentoo, Redhat, or any number of other systems extensively know that both Macports and Fink are very weak by comparison.

Macports and Fink aren't very week, they're a joke. I'd love to see Apple step up and start their own ports system, but I doubt they will (not pretty/flashy like Expose/Core Animation/Spotlight).

Originally Posted by besson3c 

3) As stated eloquently here virtualization is all the rage. OS X server is not supported as a guest OS, and it makes a crummy host OS with the products currently available too.

OS X Server can be virtualized with Parallels Server on OS X Server hosts. It's a pretty weak setup, but it is available.

Originally Posted by besson3c 

5) Apple needs to either innovate in this area, or else revamp their whole open source philosophy and commitment to work better within the OSS community

Has their philosophy changed from "dump giant tarballs of code with no changelogs"? That's the last I knew.

Originally Posted by besson3c 

7) Software Update is brain-dead and needs to be re-thought for this area. An admin running a web or database server does not care about Quicktime updates, and definitely doesn't want to have to plan to restart the server after these updates have been applied. Chances are, they don't care about iPod software updates either. One fix here is to provide the option to not receive notifications for these things, but for point updates Apple may want to look at ways to prevent having to restart the machine unless the kernel has been modified, or else break up the point updates into smaller pieces allowing admins to only install the pieces of these that they are interested in.

They really need to break the updates up for servers. "10.5.2" may work with clients, but server admins don't want to upgrade the IMAP, DHCP, and DNS servers just to get the latest Apache (without building from source, which is painful, because of #2).

[besson3c]

Originally Posted by mduell 

Has their philosophy changed from "dump giant tarballs of code with no changelogs"? That's the last I knew.

My general impression is that this hasn't changed, and they also don't provide any documentation for testing either. Their open source strategy seems to me more like an effort to get their stuff to install on other OSes (which often only happens through the work of others) than an effort to actually encourage participation and development. I also get a vibe of there being a lot of secrecy, much like Apple has for all of their other products. All of this makes not wanting to devote time to something highly secretive and tightly controlled understandable, although perhaps it is a good way to get yourself a job working for Apple?

They really need to break the updates up for servers. "10.5.2" may work with clients, but server admins don't want to upgrade the IMAP, DHCP, and DNS servers just to get the latest Apache (without building from source, which is painful, because of #2).

Yes, agreed.

I also forgot to add that if you install your own open source pieces (say a replacement to Apple's Apache, which is often several versions out-of-date, and I don't think they document security/bug fixes that have been backported), expect some conflicts, and expect to give up using Apple's GUI. Macports and Fink do not want to build against the libraries that Apple has included, understandably so, so often times this means a lot of redundancy packages. All of this could be solved if Apple would build their own ports system!

[Chuckit]

1. Actually, you can control screen sharing from the command line. Mac OS X Hints has the skinny.

2. Honestly, who cares if Apple "encourages participation and development"? I thought the point of open-source software was to allow people free access to your code and the ability to do with the what they will, not to give FOSS coders backrubs. What is the benefit to Apple doing more?

[besson3c]

Originally Posted by Chuckit 

1. Actually, you can control screen sharing from the command line. Mac OS X Hints has the skinny.

2. Honestly, who cares if Apple "encourages participation and development"? I thought the point of open-source software was to allow people free access to your code and the ability to do with the what they will, not to give FOSS coders backrubs. What is the benefit to Apple doing more?

If I were a kernel developer and did some stuff only to have it break after the next OS update, only to have Apple release a tarball several months layer with a bunch of undocumented changes, I probably wouldn't be encouraged to use Apple's code unless there were no other options.

Part of opening up source code is to give others the privilege of using your code, but the other important part that Apple is missing out on is that it allows for others to make contributions and improve your product, allowing you to reap the benefits without having to put in this same time and effort. It goes both ways, that is the benefit to Apple.

[hmurchison2001]

I think Apple's lack of focus is coming from the top of the company. Apple has what 17 thousand employees according to Wikipedia? That's a LOT of people.

The reality is Apple rose to prominence based on the wild success of the iPod which filled the company cofffers nicely until OS X could finally become the OS we knew it could.

I love the iPhone success and the iPod success but with nigh 20k employees and the fact that their hardware is outsourced I expect that Apple be able to leverage OS X a little better at the biz level. Though when you think about it...Steve Jobs has never been able to sell to business or education. He left Apple and founded NeXT and couldn't get business or education to purchase. Now he's back at Apple and displaying the same penchance for not "getting" business or education (witnessed by the PowerSchool acquisition that went nowhere)

If Apple is going to have success in some other verticals a reorg is going to be necessary. Steve Jobs' charisma has carried them foreward to great success but I don't think he can break through without splitting the company into focus areas and have capable leadership for each of these Business Units.

There needs to be someone accountable for improving the business reach of Apple. Someone accountable for portable devices and consumer electronics and they need to effectively discuss and share resources.

Apple needs to become the company that is not only envious because of their products but how they efficiently run their company.

[peeb]

I don't think Apple is that interested in enterprise computing. They see it as low margin lowest common denominator box-moving. I know the "i'm a mac' ads are not aimed at enterprise, but it gives you some insight into what Apple is thinking.

[besson3c]

hmurchison: I agree...

I've suggested some of this in here before (although never suggesting a re-org), and some people replied by saying "well, what Apple is doing now is working for them, look at their stock!" or that there is no great need for Apple to go after business because they are doing well where they are...

This is true, Apple is indeed doing well right now, but ask any marketing expert and they will tell you that it is unwise for a company to just sit on a product forever. Right now, the iPod may still be considered a "star" for Apple, but it will eventually become a cash cow, and you cannot grow your company by sitting on cash cows forever. Every company does need to grow if they want to remain relevant and not pushed aside by their competition. Instead, Apple needs to be looking at more stars. They can do so either by moving into markets that few have tapped into, or by competing directly against their competitors in existing markets.

It may be that Apple does not want to compete directly against Microsoft and Red Hat and all the others, but I would argue that there is a ton of money and business to be had there, and that they could do quite well. They have a great amount of capital and talent from within. They already sell great hardware. Doing so may help sell more iPhones and more Mac workstations. Doing so may get more Macs into the consumer markets too. In my opinion, this might be Apple's next frontier... If they do expand, they just need to be careful to not lose their focus, but with the right strategy they can make this work for them.

[besson3c]

Also, the fact that you cannot sit on a cash cow is reflected by Microsoft. I mean, they are an extremely wealthy company... Why not just sit on their Windows and Office products forever? Why develop things like the Zune, XBox, their Live/online stuff if what they have is making great money? Why risk looking bad and failing with products like the Zune?

[mduell]

Originally Posted by besson3c 

Why risk looking bad and failing with products like the Zune?

Huh?

Macintosh comes up sixth in personal computer sales and it's a raging success.
OS X is third in the OS market and it's a raging success.
iPhone barely takes third in the smartphone market and it's a raging success.
Zune solidly takes third place in the PMP market and it's a failure that looks bad.

That's quite the slanted view you've got there.

[Chuckit]

Apple is developing new things all the time. Just because they're not going for one particular market doesn't mean they aren't. Actually, I'm pretty sure they released a remarkably successful new product less than a year ago.

Also, what is this about Apple not getting education? College students are eating Apple products up with a freaking spoon.

[Chuckit]

Originally Posted by mduell 

OS X is third in the OS market and it's a raging success.

"The OS market"? Be serious here. OS X isn't really competing with POS operating systems or corporate servers. In the desktop computing market, which is where it's primarily aimed, OS X is second and gaining pretty quickly.

[besson3c]

Originally Posted by mduell 

Huh?

Macintosh comes up sixth in personal computer sales and it's a raging success.
OS X is third in the OS market and it's a raging success.
iPhone barely takes third in the smartphone market and it's a raging success.
Zune solidly takes third place in the PMP market and it's a failure that looks bad.

That's quite the slanted view you've got there.

Huh? I'm confused... What do you mean?

[besson3c]

Originally Posted by Chuckit 

Apple is developing new things all the time. Just because they're not going for one particular market doesn't mean they aren't. Actually, I'm pretty sure they released a remarkably successful new product less than a year ago.

If this is in response to me, I was referring to the enterprise sector where they are virtually a non-player now...

Apple needs to either innovate in this area, or else revamp their whole open source philosophy and commitment to work better within the OSS community

[Chuckit]

Originally Posted by besson3c 

If this is in response to me, I was referring to the enterprise sector where they are virtually a non-player now...

Yes, and I was referring to the enterprise sector when I said "Just because they're not going for one particular market…". I just don't see why every company on earth needs to aim for domination of that particular market. There are plenty of other houses serving the enterprise's needs quite well, I think.

And again, why does Apple need to revamp its open-source philosophy?

[besson3c]

Chuckit: I responded to that question above. How about we use that as a starting place?

Can I also ask why you seem to be so quick to defend Apple? Does it irritate you when people say critical things of Apple? I don't mean this question in a critical way, I'm just wondering if this is personal to you in some way...

[Chuckit]

Originally Posted by besson3c 

Chuckit: I responded to that question above. How about we use that as a starting place?

If you're referring to your "Well, maybe some Joe Schmoe working on the kernel code in his mom's basement would be upset when the kernel code changes in a way that breaks whatever he was working on and Apple isn't giving him the support he needs to figure out where and he can't be bothered to get an ADC support ticket." Well, you know, that's a valid concern for Joe — but obviously it doesn't matter all that much to Apple. Apple has its own kernel development team that it depends on, and anything coming from outside is just gravy. Is it worth the time it takes to sit down and hold all the gravy-makers' hands for free? Apple apparently doesn't think so, and I'm not convinced Apple is wrong.

It's not that I have anything against people working on open-source software, and I wish them all the best, but I don't think independent OSS coders are necessarily the end-all-be-all of software development.

Originally Posted by besson3c 

Can I also ask why you seem to be so quick to defend Apple? Does it irritate you when people say critical things of Apple? I don't mean this question in a critical way, I'm just wondering if this is personal to you in some way...

It's not that I particularly care when people criticize Apple. It's more that it irritates me when people insist on telling us how important "the enterprise" or OSS is. Apple is doing amazingly well and expanding into new markets all the time, but somehow because it's not giving somebody's pet market enough attention, it's doomed to fail or already dead or some other such nonsense.

There are already tons of companies sucking on the enterprise teat, none of them benefit me at all, and it doesn't seem like Apple is going to improve the world significantly by becoming more like them. Meanwhile, there are lots and lots of other areas where people also use computers and software, and there are often times when even "enterprise" customers need that sort of thing. I think it's perfectly fine to focus your efforts on the places where you can excel.

[besson3c]

Originally Posted by Chuckit 

If you're referring to your "Well, maybe some Joe Schmoe working on the kernel code in his mom's basement would be upset when the kernel code changes in a way that breaks whatever he was working on and Apple isn't giving him the support he needs to figure out where and he can't be bothered to get an ADC support ticket." Well, you know, that's a valid concern for Joe — but obviously it doesn't matter all that much to Apple. Apple has its own kernel development team that it depends on, and anything coming from outside is just gravy. Is it worth the time it takes to sit down and hold all the gravy-makers' hands for free? Apple apparently doesn't think so, and I'm not convinced Apple is wrong.

It's not that I have anything against people working on open-source software, and I wish them all the best, but I don't think independent OSS coders are necessarily the end-all-be-all of software development.

If Apple or whomever doesn't feel that there are worthy developers out there, fine, your argument make sense, but conveniently I used the kernel as an example... In the case of the kernel, you can bet your ass that there are great developers out there that can help Apple with it. For starters, a good percentage of the code base belongs to open source in the first place. Additionally, there is all sorts of kernel development going on in developing certain features, driver support, optimization, security fixes, etc. There are probably many more OSS developers working on low level stuff like this than there are GUIs and consumer apps.

It's not that I particularly care when people criticize Apple. It's more that it irritates me when people insist on telling us how important "the enterprise" or OSS is. Apple is doing amazingly well and expanding into new markets all the time, but somehow because it's not giving somebody's pet market enough attention, it's doomed to fail or already dead or some other such nonsense.

There are already tons of companies sucking on the enterprise teat, none of them benefit me at all, and it doesn't seem like Apple is going to improve the world significantly by becoming more like them. Meanwhile, there are lots and lots of other areas where people also use computers and software, and there are often times when even "enterprise" customers need that sort of thing. I think it's perfectly fine to focus your efforts on the places where you can excel.

So, you'd prefer Apple to grow and expand outside of business, the iPhone 2.0 stuff notwithstanding, and just leave business to MS and Linux/Unix? You don't think Apple will ever want to sell hardware to the business sector?

[mduell]

Originally Posted by besson3c 

So, you'd prefer Apple to grow and expand outside of business, the iPhone 2.0 stuff notwithstanding, and just leave business to MS and Linux/Unix? You don't think Apple will ever want to sell hardware to the business sector?

Apple likes high margin businesses; most of the enterprise IT markets where they could compete in the neat future are not high margin.

[besson3c]

Enterprise software is very high margin, as are products such as SANs... Are you thinking in terms of enterprise level workstations?

[P]

Very interesting discussion. I just want to note a few things: Apple very rarely does things half way. It's the best in the industry, or it's nothing. Mac OS X Server is an exception - it's certainly not the best, yet it hasn't been killed. I think that Apple recognizes the need to have something, but they don't have the management bandwidth to make it best in class.

The solution, I think, is to partner with some server vendor to make existing server products work better with Macs. Sun is a possibility, except they've tried that before and failed miserably. No, the best fit is probably to partner with IBM. Those persistent rumors about IBM buying Macs seem to point in that direction, in my opinion.

Also note that Leopard has been certified to as being compliant with the UNIX 03 specification. I doubt they'd bother with doing that unless they had some sort of plan with it.

[P]

Originally Posted by besson3c 

If Apple or whomever doesn't feel that there are worthy developers out there, fine, your argument make sense, but conveniently I used the kernel as an example... In the case of the kernel, you can bet your ass that there are great developers out there that can help Apple with it.

In which case I suggest they apply for a job at Apple. If they're good enough, I think they'll get hired.

I really don't think the kernel is the best place to ask for outside help. It is the area most in need of tight management, and considering Apple's culture of secrecy, I doubt that that sort of management would be conducted on a public mailing list.

Originally Posted by besson3c 

For starters, a good percentage of the code base belongs to open source in the first place.

Yes, all of it. See, the kernel is open source. If you meant to say that it comes from open source development to begin with, then no. Sorry. It comes from the Mach research project at Carnegie Mellon University, then further developed at NeXT and at Apple.

Originally Posted by besson3c 

Additionally, there is all sorts of kernel development going on in developing certain features, driver support, optimization, security fixes, etc.

Driver support is the one area where it might make sense - fortunately, there is a way to do that for XNU (the Mac OS X kernel, for those who didn't know). You can write kernel extensions. For older versions of it, the API for that was not stable, but since 10.4 it is. At least theoretically you don't need the kernel source code (although it certainly doesn't hurt.

Originally Posted by besson3c 

There are probably many more OSS developers working on low level stuff like this than there are GUIs and consumer apps.

Considering the failure (so far) of desktop Linux, I'm inclined to agree. That is not necessarily a good thing, though.

[besson3c]

Originally Posted by P 

In which case I suggest they apply for a job at Apple. If they're good enough, I think they'll get hired.

That's assuming that they want a job at Apple or don't already have a job they like, or family keeping them in a particular location.

I really don't think the kernel is the best place to ask for outside help. It is the area most in need of tight management, and considering Apple's culture of secrecy, I doubt that that sort of management would be conducted on a public mailing list.

Maybe, maybe not... I'm just saying that purely from a technical standpoint, there are people out there that can help Apple with the kernel in particular. There have been several commercial products in the history of computing that have benefited from open source development/collaboration.

Yes, all of it. See, the kernel is open source. If you meant to say that it comes from open source development to begin with, then no. Sorry. It comes from the Mach research project at Carnegie Mellon University, then further developed at NeXT and at Apple.

Parts of Apple's kernel are their pieces, some are open source (e.g. the FreeBSD team). For instance, up until Leopard OS X used ipfw as its firewall which required kernel support. One of the Apple pieces is the bit that prevents OS X from being installed on generic PCs, I believe.

Considering the failure (so far) of desktop Linux, I'm inclined to agree. That is not necessarily a good thing, though.

The failure in desktop Linux in part involves its focus and goals. Getting an OS to work with such a wide range of hardware is very challenging, and then there is the issue of usability design and who the target audience of the OS is, who the features are designed for, what the features should be, etc.

A lot of criticism can be given to desktop Linux, but its kernel is not one of these criticisms. Honestly, I would say that the low level parts of Linux in many ways are better than OS X's. For one, Linux is more responsive all-round, its window manager being separated from the rest of the OS makes it easier to run on older hardware and consume less resources, and Linux supports many technologies that OS X doesn't such as support for numerous file systems that are superior to HFS+, jails, LVM, KVM/Xen/libvirt, etc. In fairness, OS X has Linux beat in management of multiple displays, monitor detection, printing, other multimedia stuff, etc.

My point here is simply that the high level Linux stuff is what is weakest - the low level stuff does quite well.

[besson3c]

I suspect that a big reason why there is a divide between Apple and the open source community is that OS X is less modular than Linux is in many respects. Many pieces of OS X are tethered to OS X, and Apple wants it this way, for better or worse. OS X is tethered to Aqua, HFS+, etc.

I'm not saying that this is a bad thing necessarily, but leaving Apple's need for secrecy aside, this may render large parts of OS X useless to anybody wanting to incorporate or add to these technologies. There is also the risk of a company like MS flat out copying parts...

Still, there are pieces to OS X that do make sense to leave open, and its in these areas where Apple seems to have struggled with striking the right balance among all of these concerns.

[mduell]

Originally Posted by P 

Also note that Leopard has been certified to as being compliant with the UNIX 03 specification. I doubt they'd bother with doing that unless they had some sort of plan with it.

Only on Intel; I think that's just to get a check mark on various acquisition RFP/RFQs.

[P]

One thing that always annoyed me with desktop Linuxen is the scheduler, and that IS a kernel part. There was a patchset that installed a much better scheduler, but it is no longer being updated. I heard that work was progressing on a new scheduler, but I stopped caring about desktop Linux a while back, and never heard what happened.

In general I agree that the low-level UNIX stuff is better on Linux than OS X, though.

[P]

(Leopard being UNIX 03 certified)

Originally Posted by mduell 

Only on Intel; I think that's just to get a check mark on various acquisition RFP/RFQs.

Precisely my point. Private customers rarely write RFQs, so Apple does care a little about the enterprise after all.

OK, so universities DO write those things on occasion (I should know, I've been involved in them), and that IS an Apple market, but still...

[dharknes]

Originally Posted by besson3c 

1) Provide some sort of incentive to run OS X Server at all. In big enterprise environments, the just-add-water design approach just doesn't work. Each enterprise has unique approaches to account creation/maintenance, authentication, and in many cases this involves sharing data with other groups. OS X Server is designed primarily to make manual modifications to configurations easy, but these GUIs have no value to organizations needing flexible, automated workflows. In high performance computing environments, you wouldn't want the overhead of running any GUI at all.

This very much depends on what you're doing. Having the GUI "running", sitting at the login windows doesn't add any significant overhead. Using the rest of the Server Admin tools can be used remotely without the need for VNC. Or Apple had made it a point to provide every GUI function (and more) through the command. So scripting, automating, and integrating is actually pretty simple.

Originally Posted by besson3c 

2) Taking Apple's GUIs out of the picture, Apple does not offer any compelling open source package management. We can dismiss this by saying that we have Macports and Fink, but anybody that has used FreeBSD, Ubuntu, Gentoo, Redhat, or any number of other systems extensively know that both Macports and Fink are very weak by comparison.

I just started deploying a completely apple environment at a medium size university and can say that Apple doesn't provide this because its not needed. I would say apple provide 90+% of the application and services needed to run an enterprise class system. I'm in the process of replacing several Debian and RedHat linux system and at this point the only 2 item I had to install from Macports is the Nagios and Bacula clients.

One note. The Apple platform is a little different, when you start using a Mac server do NOT assume it is just another Linux, BSD, or FOSS platform. It has some unique benefits that make FOSS and other Unix system look old and crappy.

Originally Posted by besson3c 

3) As stated eloquently here virtualization is all the rage. OS X server is not supported as a guest OS, and it makes a crummy host OS with the products currently available too.

OS X Server is supported as a guest on OS X Server. The underlying reason (beyond selling more Apple HW) is that it allows the guest OS to take advantage of special hardware monitors and other hardware. But having just deployed 40 virtualized servers I will also say the virtualization is really not all that great. It offers little in performance or security and really only makes it easier to understand "where" services are running.

[QUOTE=besson3c;3661154]4) Apple would need to revamp their whole software support infrastructure to assist with supporting leaving the comfortable little Apple bubble and getting into the nitty gritty of low level software issues in a multi-platform Unix centric environment[/QUOTA]

I'm not really sure what you mean here. I had little or no problem integrating Apple into our campus LDAP (LDAP not AD) system and other unix system such as kerberos. We do automated provisioning through perl and php scripts. And to be fair the provisioning process on the Apple servers was easier then it was on our Linux system. Launchd allowed us to provide a near really time process that didn't require us to write daemons and a bunch of other custom infrastructure just for our provisioning.

And if you look real close you'll see that 98% of Apple server is OSS under the hood and Apple provides nice abstraction layers to reduce the administrative complexity. Linux and FOSS could look a lot from Apple.

Originally Posted by besson3c 

5) Apple needs to either innovate in this area, or else revamp their whole open source philosophy and commitment to work better within the OSS community

So really what you want it Apple and Mac OS server to be another Linux system. I've been a Linux user/admin since 1994 and can very confidently say that OS server is head and shoulders available "enterprise" linux distributions such as RedHat. As I stated above most of Apples system has open source roots and Apple contributes a lot back, but Apple should never become just another FOSS. Linux is nothing more than the least common denominator of computing and has introduced very little innovation.

Originally Posted by besson3c 

6) Apple needs to stop being schizophrenic in how it designs its own software. Some pieces of OS X support Unix conventions nicely, but others don't. For instance, there is no way to start/stop Apple's VNC solution or application firewall remotely without a GUI. As stated, in many environments the admins are not going to want to run the GUI at all, but are certainly not going to be interested in being dependent upon VNC for remote administration - especially when dealing with clusters of machines. Also, if Apple insists upon developing proprietary solutions over open standards, they need to provide a compelling reason for this. They did the right thing by opening up launchd (and I'm assuming it is documented too), but not so with Screen Sharing and the Application Firewall

I'm not exactly sure what you mean about the VNC system. But I do most of my OSX server admin work through ssh. You can start and stop the ScreenSharing server from the command line using launchd.

OS X server doesn't have the Application firewall. It uses IPFW which can absolutely be controlled from the command line. As a matter of fact there is really no reason to use Apple firewall management application. Other then it one of the best firewall manager I've ever used.

Originally Posted by besson3c 

7) Software Update is brain-dead and needs to be re-thought for this area. An admin running a web or database server does not care about Quicktime updates, and definitely doesn't want to have to plan to restart the server after these updates have been applied. Chances are, they don't care about iPod software updates either. One fix here is to provide the option to not receive notifications for these things, but for point updates Apple may want to look at ways to prevent having to restart the machine unless the kernel has been modified, or else break up the point updates into smaller pieces allowing admins to only install the pieces of these that they are interested in.

I'll give a 50% on this one because

1) You can turn off updates you're not interested in and you won't be notified.
2) Because of some cool features like PodCast producer QT does become an requirement for running a web server.
3) The reboot on upgrade/update/patch doesn't bother me much since I reboot my Linux and Solaris boxes when I patch as well. I do this as part of my testing to make sure everything about the system is working. In the past I've had Debian packages change the startup scripts for a server the actually prevented the service from starting or the reenabled services I disabled and I didn't catch the change for weeks because I didn't reboot the system.

[dharknes]

Originally Posted by Chuckit 

1. Actually, you can control screen sharing from the command line. Mac OS X Hints has the skinny.

2. Honestly, who cares if Apple "encourages participation and development"? I thought the point of open-source software was to allow people free access to your code and the ability to do with the what they will, not to give FOSS coders backrubs. What is the benefit to Apple doing more?

At this point the FOSS world it is rarely about code sharing and reuse. It is more about sticking it to the man and whining about how much better FOSS is over commercial development. The problem with this is most FOSS is really just a bad clone of some commercial product then the FOSS developer didn't want to pay for.

[dharknes]

Originally Posted by besson3c 

I suspect that a big reason why there is a divide between Apple and the open source community is that OS X is less modular than Linux is in many respects. Many pieces of OS X are tethered to OS X, and Apple wants it this way, for better or worse. OS X is tethered to Aqua, HFS+, etc.

From a kernel perspective maybe OSX is less modular. But the linux kernel is also based on 30 year of technology. Not start a different argument but my biggest complaint with the linux kernel is its monolithic design. Trust I'm not suggesting that Apple XNU kernel is much better but it's my hope that they will continue the work of the Mach 3 kernel and produce a full micro-kernel design.

I also don't see a lot of support for ZFS on Linux.

Originally Posted by besson3c 

I'm not saying that this is a bad thing necessarily, but leaving Apple's need for secrecy aside, this may render large parts of OS X useless to anybody wanting to incorporate or add to these technologies. There is also the risk of a company like MS flat out copying parts...

Still, there are pieces to OS X that do make sense to leave open, and its in these areas where Apple seems to have struggled with striking the right balance among all of these concerns.

Personally I like Apple approach of open sourcing or using OSS for infrastructure and building closed source value add on top.

[besson3c]

Originally Posted by dharknes 

This very much depends on what you're doing. Having the GUI "running", sitting at the login windows doesn't add any significant overhead. Using the rest of the Server Admin tools can be used remotely without the need for VNC. Or Apple had made it a point to provide every GUI function (and more) through the command. So scripting, automating, and integrating is actually pretty simple.

All of this does have a significant overhead in memory consumption, albeit much less so to idle... What if your server is memory bound?

I just started deploying a completely apple environment at a medium size university and can say that Apple doesn't provide this because its not needed. I would say apple provide 90+% of the application and services needed to run an enterprise class system. I'm in the process of replacing several Debian and RedHat linux system and at this point the only 2 item I had to install from Macports is the Nagios and Bacula clients.

It's not needed to run a bunch of computer labs, but it's definitely needed to run open source infrastructure and keeping these pieces and their dependencies up-to-date.

One note. The Apple platform is a little different, when you start using a Mac server do NOT assume it is just another Linux, BSD, or FOSS platform. It has some unique benefits that make FOSS and other Unix system look old and crappy.

Such as?

OS X Server is supported as a guest on OS X Server. The underlying reason (beyond selling more Apple HW) is that it allows the guest OS to take advantage of special hardware monitors and other hardware. But having just deployed 40 virtualized servers I will also say the virtualization is really not all that great. It offers little in performance or security and really only makes it easier to understand "where" services are running.

Huh? What is your understanding of the benefits and purpose of virtualization? It sounds like there is a disconnect here...

I'm not really sure what you mean here. I had little or no problem integrating Apple into our campus LDAP (LDAP not AD) system and other unix system such as kerberos. We do automated provisioning through perl and php scripts. And to be fair the provisioning process on the Apple servers was easier then it was on our Linux system. Launchd allowed us to provide a near really time process that didn't require us to write daemons and a bunch of other custom infrastructure just for our provisioning.

Try setting up OS X Server to function as a fully fledged mail server, a web server with a very specific configuration (say, php-cgi in a chroot jail), etc. It becomes a pain in the ass when you leave Apple's little bubble, as I said. Aside from SSL LDAP calls apparently breaking from time to time across OS X Server updates (so I've been told), getting OS X Server to work in computer labs is pretty easy, but using it for much else becomes increasingly less fun.

And if you look real close you'll see that 98% of Apple server is OSS under the hood and Apple provides nice abstraction layers to reduce the administrative complexity. Linux and FOSS could look a lot from Apple.

What could they learn? We don't use pretty GUIs in an enterprise environment very much, for good reason.

So really what you want it Apple and Mac OS server to be another Linux system. I've been a Linux user/admin since 1994 and can very confidently say that OS server is head and shoulders available "enterprise" linux distributions such as RedHat.

How so? Ports system - nope, something like the RedHat satellite - nope, access to better file system options - nope, better virtualization products - nope, better software update procedures - nope, something like LVM for managing partition/slice usage (or does OS X Server still install everything onto a single partition/slice?)... what area is OS X Server better in the enterprise? Say you have a cluster of 10 servers running a particular open source based service... Where is the advantage to running OS X Server?

As I stated above most of Apples system has open source roots and Apple contributes a lot back, but Apple should never become just another FOSS. Linux is nothing more than the least common denominator of computing and has introduced very little innovation.

I would argue that the open source community as a whole as innovated far more than Apple has to the computing world, they just innovate in completely different areas.

OS X server doesn't have the Application firewall. It uses IPFW which can absolutely be controlled from the command line. As a matter of fact there is really no reason to use Apple firewall management application. Other then it one of the best firewall manager I've ever used.

I stand corrected... I assumed that Server used the Application firewall as client does. I'm not sure what you need a firewall manager for to generate firewall rules anyway, but okay...

3) The reboot on upgrade/update/patch doesn't bother me much since I reboot my Linux and Solaris boxes when I patch as well. I do this as part of my testing to make sure everything about the system is working. In the past I've had Debian packages change the startup scripts for a server the actually prevented the service from starting or the reenabled services I disabled and I didn't catch the change for weeks because I didn't reboot the system.

So you reboot every time you apply any update whatsoever? Weird policy...

[besson3c]

Originally Posted by dharknes 

At this point the FOSS world it is rarely about code sharing and reuse. It is more about sticking it to the man and whining about how much better FOSS is over commercial development. The problem with this is most FOSS is really just a bad clone of some commercial product then the FOSS developer didn't want to pay for.

Really? Please find me a commercial counterpart that is better than the following:

- Apache
- Bind
- Postfix/Sendmail
- Cyrus IMAP
- SNMP/RRD/Cacti
- MIT-Kerberos
- Python/Ruby/Perl
- emacs
- Mailman
- OpenSSL
- OpenLDAP
- Subversion
- OpenSSH
- The BSD/Linux kernels

How is it that on one hand you can go on about how great OS X Server is, which is absolutely dependent on many of these technologies (not to mention its kernel), yet be dismissive of open source as being a bad clone?

What you meant to say that OSS GUI desktop apps are often bad clones, but you didn't qualify what you said this way.

[besson3c]

Originally Posted by dharknes 

From a kernel perspective maybe OSX is less modular. But the linux kernel is also based on 30 year of technology. Not start a different argument but my biggest complaint with the linux kernel is its monolithic design. Trust I'm not suggesting that Apple XNU kernel is much better but it's my hope that they will continue the work of the Mach 3 kernel and produce a full micro-kernel design.

I also don't see a lot of support for ZFS on Linux.

Huh? Are you trying to get into the nitty gritty of mach vs. micro kernels? You do realize that the differences are very esoteric, right? The Linux kernel does support kernel extensions, and you can control what gets included in the kernel just like you can in a BSD kernel.

You don't see a lot of support for ZFS anywhere yet, do you? You can get experimental support in FreeBSD 7, and I guess some developer build of OS X or something... What's your point?

[P]

ZFS support is included in Leopard, but read-only by default. Read-write, non-bootable is available from semi-official macosforge.

Plug-in filesystem support OUGHT to be a big advantage of the XNU design but that hasn't materialized - probably because it was tricky to test before the kernel API was frozen in 10.4. The only third-party plug I'm aware of is the e2fs one. IMO, FUSE makes that less relevant anyway (although you could argue that moving the filesystems into userspace is consistent with more modern microkernel thinking).

[Peter]

Apple doesnt want the enterprise market, though.