Apple Computer is guarding its server plans closely, leaving observers to wonder what will come of the product line. The Mac manufacturer said on April 5 that it will preview its new Mac OS X 10.5, dubbed Leopard, at its Worldwide Developers Conference in August. But the company has yet to reference a Leopard server edition. Nor has it hinted about any plans it might have to move its Xserve server line to Intel processors from the PowerPC under its June 2005 decree to move from PowerPC chips to Intel processors by June 2007. On a related note, another Apple product’s future is uncertain. Apple has asked the development team behind Aperture to leave— probably related to the fact that Aperture got some pretty harsh reviews after its release.
Honestly Thom, does presenting rumours from ThinkSecret really count as OS news that the Aperture team have been disbanded?
To me, Yes.
Think Secret has probed reliability in its articles.
Yes, but then Apple clamped down on leaks, its sources dried up, and it’s been nothing but baseless supposition ever since. Most of its predictions in the last 12 months have not come to pass.
As for Aperture’s future, it’s assured. There is clearly a market for that kind of a product, and currently Apple is out the door early and still has some time to get a leg up on Lightroom.
It’s more likely that they’re re-organising the project team, bringing in some new people, and weeding out the ones they thought were bad, so that version 2.0 will be that bit better. This isn’t anything new for Apple, Final Cut Pro was pretty poor when it was first released, but then got gradually better, and now it’s one of the most popular apps in its field.
As for Mac OS X Server, they’ve never hightlighted it. There was no mention of OS X.4 Server at any of the keynotes, but it was released nonetheless, and it’s going to be the same with Leopard. As for the XServes, they’re waiting on Intel for either Conroe (August) or Woodcrest (January/February). Woodcrest would probably make more sense, but they might stick a Conroe chip in and offer a new configuration in the new year with Woodcrest.
While the server business isn’t huge for them, there is a certain prestige to having it, and there would be a certain loss of face in ditching it. Certainly http://www.army.mil wouldn’t be happy.
Edited 2006-04-27 23:39
You look like someone who just can’t acept Apple can blow it sometimes, just take a look to Aperture, Apple is cutting the price like Think Secret said.
Edited 2006-04-28 00:17
because they cut the price the product “blows”?
have you actually used it?
I never said Aperture was good, I said there was a good market for it. And by all accounts, sales have actually been pretty good, despite all the warts. Thinksecret didn’t say they were cutting the price (Apple did that six weeks ago when Aperture 1.1 came out), it said they had fired the entire engineering team, which is, frankly, unlikely.
They’ve been a 50/50 crap shoot over the past year.
Apple and thats desktops, ipods and DRM.IF I had a Mac it would be wasted as a server thats for sure.
I run a MacMini server and Microsoft run over 150 MacMinis as servers, not counting the numerous G3s, G4s and G5s they have as servers.
So, do you work here (http://davidweiss.blogspot.com/2006/04/tour-of-microsofts-mac-lab.h…)
They don’t.
The blog post states that they use the Mac Minis to run automated tests on software builds. The machines doing the serving is Xserve racks.
Yes, because you have just no clues about Mac OS X Server technologies.
There server platform is pretty good, far easier to setup compared to other systems.
It’s a Shame about Aperture, I’ve used and quite liked it, some parts were a little clunky, and was pretty slow doing some things, but overall it wasn’t too bad.
“Pretty good” if all you do is run a small workgroup-style network, and don’t want to do anything more than basic webhosting and such.
Anything more than that requires dropping to the commandline, and editing config files just like on any other unix. Not only that, but you get osx’s horrid server performance to boot!
Wake me up when Apple lobotomizes OSX and sticks a good performing kernel in there for server tasks. Until then, I’ll keep OSX around on my desktop, but it will never see the light of day on a server.
I worked for Ticketmaster a while back and they did miracles with small machines. If you want good server performance for your database applications, write your own system and don’t use off-the-shelf packages. Using SQL often leads to tabelizing data structures which aren’t naturally 2 dimensional. Prefetching disk blocks before they are asked-for is only possible when the server software knows how to anticipate based on the nature of the database.
f you want good server performance for your database applications, write your own system and don’t use off-the-shelf packages.
I don’t consider something like PostgreSQL an “off-the-shelf” package.
Prefetching disk blocks before they are asked-for is only possible when the server software knows how to anticipate based on the nature of the database.
Oh sure, and everyone who is writing a business application wants to take the time to reimplement all the old classic low-level problems of data management, such as locking, concurrency, transactions, domain constraints, etc… While you’re at it, why not just build your own dedicated OS, too?
The best separation of concerns is to let a robust, proven system handle the low-level data management problems. By the time you implement all the safeguards a serious business database system needs, you will find you have reinvented the RDBMS wheel.
Ticketmaster did build an OS. So did I. http://www.justrighteous.org it’s called LoseThos. Obviously such a feat would only benefit those with very critical performance demands, but the Apple thing I was responding to was a guy mentioning performance.
I think Ticketmaster is switching to Linux, though.
Anything more than that requires dropping to the commandline, and editing config files just like on any other unix. Not only that, but you get osx’s horrid server performance to boot!
Like any good modern unix OSX offers the choice between ordinary traditional flat files and retrieveing data from dedicated services that probably store data in some sort of data base or propriatary format. This is a feature, just ask any sysadmin worth is salt (flat files are nothing less than a blessing in most cases, especially when diagnosing problems).
The horrid server performance doesn’t need to be an issue if you code specifically for OSX. IIRC the original problem was detected with a MySQL benchmark (highlighting the problem of a slow fork() function). However a version of Oracle and Sybase coded specialy for OSX manages to avoid hs issue.
Dixit Anandtech : “When we asked Apple for a reaction, they told us that some database vendors, Sybase and Oracle, have found a way around the threading problems.”
That’s wonderful, but I’m basing my performance appraisel on my own experience, not on that terrible anandtech review. OSX Server does not cut it for me.
It *is* an issue for my usage patterns.
My point about the editing of files was – the OPs point about the GUI tools for configuration etc, making the OS “easy” to administrate as a server. That doesn’t hold true once you go beyond simple changes, that was my point and I still stand by it. I’m not arguing flat files/services for configuration.
Obviously, with my Solaris/FreeBSD servers, it’s either flat file or ldap for me.
I just don’t see any *advantage* to running OSX Server for anything more than a workgroup server. That was my point (and the reason I don’t use OSX Server, only client…)
Cheers.
PS – “if you code specifically for OSX” – no thanks. I’ve had enough vendor lockin for a few lifetimes. That’s *never* a good solution to a problem.
Isn’t that why Linux even exists today? 
PS – “if you code specifically for OSX” – no thanks. I’ve had enough vendor lockin for a few lifetimes. That’s *never* a good solution to a problem.
Isn’t that why Linux even exists today? 
I don’t see any problems with optimizing a database or any other kind of server software for optimal performance on an OS. MySQL doesn’t optimise for OSX and it performs terribly, Oracle does and doesn’t suffer the same penalty.
My problem is with porting over code to OsX that has been tweaked for performance for years on Linux and Windows and then complaining it runs like a dog. For better or worse Apple made something unique with XNU.
“I just don’t see any *advantage* to running OSX Server for anything more than a workgroup server.”
WebObjects ?
Video Streaming ?
Xgrid ?
Windows integration ?
Xsan ?
spotlight module for apache ?
WebObjects – *laughs* you must never have done major design work, or you just went to apple.com to get your information. Other than apple.com, I’ve never seen a large scale deployment of “WebObjects” and I certainly can name a lot of other technology that can do the same thing.
Video Streaming – *any os*. I suppose you meant quicktime video streaming? Who cares, vendor lockin issue again.
Xgrid – I found it pretty damn easy to write MPI aware software. I don’t really see how Xgrid makes it any easier.
Windows integration – Other than at the workgroup level, I don’t see how OSX does this any different than anything else. You do realize OSX is using Samba, right? So can FBSD/Linux/etc
Xsan – I’ll give you this one, it’s a pretty nice tool. Can’t argue.
Spotlight module for apache – sucks/pointless. Can do it all with other methods, and be more efficient at it.
Good try, but even Xsan is doable with less expensive solutions. Xsan is certainly easier to deal with though, so you get +1 for that. -5 for the rest though, beyond workgroup space you’re doing the same amount of work to get that stuff functional (if not more) on OSX as you would on FBSD/Linux/Solaris/etc.
–WebObjects – I’ve never seen a large scale deployment of “WebObjects”
iTunes, .Mac, Apple Store, BBC, Deutsche Bank, DisneyLand, Walt Disney World, United States Postal Service , etc …
But I think that for you, a store like iTunes, selling more than 1 billions of songs isn’t a large scale deployment …
–you must never have done major design work
You must never have used Webobjects nor even know anything about it …
–I certainly can name a lot of other technology that can do the same thing.
I highly doubt this.
–Video Streaming – *any os*. I suppose you meant quicktime video streaming? Who cares, vendor lockin issue again.
Because streaming MP3, AAC, MPEG-4 and 3GPP is a vendor lockin for you ?
Is isn’t because there is “Quicktime” in the name, that only quicktime files can be streamed. If you had just read a basic documentation about it you will know that.
Anyway, vendor lockin issue is a political issue, it is YOUR issue, not a Mac OS X one.
–Windows integration – Other than at the workgroup level, I don’t see how OSX does this any different than anything else. You do realize OSX is using Samba, right? So can FBSD/Linux/etc
Perhaps with compatible windows 2003 ACL integrated in Tiger ? There is also WorkGroup Manager despite the fact you don’t want to count it for Mac OS X.
It is so easy to say: “Other than all good stuffs built in Mac OS X, Mac OS X suxes …”
–Spotlight module for apache – sucks/pointless
Tell me how your method can be more efficient than an indexed search built in the OS ?
Again, your post only prooves that you don’t have any clues about macosx server. Your are only prooving that Mac OS X is able to handle all the things that linux/fsbsd can.
From this, I just don’t see any *advantage* to running Linux/FBSD when I already have Mac OS X. (Solaris 10 is out of this scope as I consider it as the best OS server, far away Mac OS X, Linux and FBSD).
— -5 for the rest though
Wow, -5 to Mac OS X just because it can do the same that Linux/FBSD, what a fair point of view.
–beyond workgroup space you’re doing the same amount of work to get that stuff functional (if not more)
Only if you don’t know Mac OS X.
Maybe you should learn to quote me correctly. You can’t just snip out parts of a comment which nullify what you are saying just so you *sound* like you have a legitmate argument. That’s bad form.
“Other than apple.com, I’ve never seen a large scale deployment of “WebObjects””
Guess what, the three big ones (first in your list) you listed are part of Apple. *I* can create a web-language and use it that will work for my needs. I can also throw enough hardware at it to make it work well.
As for the others:
bbc.co.uk runs Solaris (9).
db.com was for obvious reasons unable
disneyland runs IIS (windows)
usps runs netscape enterprise server, this isn’t available for OSX.
*ding ding* sorry, you lose.
“You must never have used Webobjects nor even know anything about it”
Actually I do, a lot more than just reading the WEBPAGE which is all you apparently did. You didn’t even do a good job of that:
“Platforms
* Mac OS X v10.4.1with Xcode 2.1 (development)
* Mac OS X Server v10.4.1
”
Aww, looks like all of the places you listed cannot possibly run Webobjects (it is non-supported on any other OS – even though it’s written in java) on their frontend, nor even OSX. I wonder why. No big business is going to run unsupported tech, sorry.
“I highly doubt this. ”
Doubt no more!
1. SOPE
2. GNUstepWeb
3. Apache Tapestry
4. Cayenne
All from the same wikipedia page you probably gleamed your information from. At least mine can be verified, yours is wishy-washy. (http://en.wikipedia.org/wiki/WebObjects)
“Because streaming MP3, AAC, MPEG-4 and 3GPP is a vendor lockin for you ?
Is isn’t because there is “Quicktime” in the name, that only quicktime files can be streamed. If you had just read a basic documentation about it you will know that. ”
No, you snide ass, the only video streaming OSX does better/that cannot be done on another os IS Quicktime. All of the others I can stream from other platforms, with BETTER PERFORMANCE. Which was my whole point. I really ditest your attitude towards me. Read the line you quoted from me, as I said *any* os can do streaming, only quicktime is limited to OSX….
“Anyway, vendor lockin issue is a political issue, it is YOUR issue, not a Mac OS X one. ”
Uh?? Do you live on Mars?? vendor lockin IS NOT a political issue, it’s a MONETARY issue. Example: State of Hawaii uses OS/390 mainframes still running OS/VS COBOL II to handle all finances. Guess why? IBM Vendor lockin, they can’t convert. You have ANY idea how much that costs them? You don’t want to know…
“Perhaps with compatible windows 2003 ACL integrated in Tiger ? There is also WorkGroup Manager despite the fact you don’t want to count it for Mac OS X.”
?? To quote one of my original posts: “I just don’t see any *advantage* to running OSX Server for anything more than a workgroup server”
Why on earth are you telling me it’s a good workgroup server and that I am clueless? Are you purposefully blind and ignorant?
“It is so easy to say: “Other than all good stuffs built in Mac OS X, Mac OS X suxes …” ”
That’s nice, but other than as a workgroup server, OSX does not fare well for me, nor many others. That’s why it’s not used! Not only that, but don’t make me out to be an idiot who would use the term “suxes”. I’m not 14.
“Tell me how your method can be more efficient than an indexed search built in the OS ? ”
What the hell is the point of that? I don’t need an indexed search built into my OS, all of my webapps index themselves into a database as well, JUST LIKE SPOTLIGHT DOES FOR THE FILESYSTEM except mine do it on the logical level, which allows them to index FAR MORE than spotlight. Spotlight can only index filenames, meta information, and content of a few kinds of files. My websites can index based on user tagging, input, comments, etc. Sorry, spotlight can’t do that! Spotlight is great on OSX client, but it’s pointless on Server. Again, in one of my original posts I make it quite clear I use and like OSX as a client.
“Again, your post only prooves that you don’t have any clues about macosx server. Your are only prooving that Mac OS X is able to handle all the things that linux/fsbsd can. ”
What? I never said OSX *can’t do things*. I said other OSs can do the majority of people’s server workload more efficiently/quicker/however you want to say it. Sorry to destroy your dreams, but it’s true.
“From this, I just don’t see any *advantage* to running Linux/FBSD when I already have Mac OS X. (Solaris 10 is out of this scope as I consider it as the best OS server, far away Mac OS X, Linux and FBSD).”
#1 – Guess what server OS I run. Solaris 10. Guess what else, I migrated FROM OSX SERVER to FreeBSD due to performance issues on my large scale (1000+ server) deployment. I *do* know what I am talking about. Visit my website if you think I’m full of poo.
“Wow, -5 to Mac OS X just because it can do the same that Linux/FBSD, what a fair point of view. ”
-5 because it does it terribly/slowly/etc. Why spend all that money on hardware/OS when I can pick up a Sun server for much less and have MUCH more hardware power, AND a much more efficient OS for serving?
“Only if you don’t know Mac OS X.”
What the hell are you talking about? You try getting spam filtering working when you have virtual domains. Guess what? You’ll be editing postfix config files JUST LIKE any FreeBSD admin/Linux admin/etc. This is very BASIC stuff. Don’t even get me started on configuring mySQL properly or trying to setup postgresql or tuning apache, etc. ALL of those require hand editing config files. It’s the same as running them on any other unix, why not run them on unixes that are much more efficient at serving and run on much less expensive (but faster) platforms?
Sorry, but you really had it coming with that piss-poor attitude of yours. When you’ve graduated HS come back and try again.
The smartest thing Apple could do is move their server OS kernel sharply in the direction of FreeBSD, since they are already using FreeBSD for userland and system services. In fact, there is nothing stopping them from just *using* FreeBSD as the kernel, providing a nice suite of management tools on top of that (which they can keep closed-source if they want). The BSD license is totally down with that ;-).
Personally, I would consider it to be a great move for Apple and a great thing for FreeBSD, too.
First off, no, they’re not using FreeBSD for userland and system services. Darwin’s userland and a lot of its system services are inherited from newer BSDs and there’s still a fair bit of code exchange, but they are not the same thing. But having split off from mainline BSD back in the mid-1980s when NeXT first started writing the OS, it has less to do with FreeBSD than OpenBSD does. Seriously, if Apple were using FreeBSD as the basis for their OS, don’t you think there’d have been more serious progress on the PPC port of FreeBSD?
Second, there is plenty stopping them from just using FreeBSD’s kernel. They’d have to root out all the code that interfaces directly with the Mach portion of the kernel and refit it with BSD calls. That’s already no small task given that OS X is a 20-year-old operating system. Then they’d likely have to replace a whole mess of drivers. They’d also have to replace the entire I/O kit, which resides in Mach-land.
First off, no, they’re not using FreeBSD for userland and system services.
Yes I understand they are not *directly* using FreeBSD, but as you say they are using many aspects of it, such as VFS, network stack and large portions of BSD userspace. Thus, it is closer to FreeBSD than to any other known system.
And secondly, sigh… of course the system calls and I/O kit are not the same. Still, it would seem to be a shorter jump to FreeBSD than, say, dropping some other completely new kernel in there. As I understand, there is talk of Apple doing away with the microkernel. What do you think they are planning on? I am of course not a Mac expert*, so I would be interested in your guess.
*(I had a short relationship with XServe a couple years ago as I tried to help someone deploy a major PHP/MySQL web application on a Mac server, and it was painful. XServe definitely has major performance and scalability problems.)
As I understand, there is talk of Apple doing away with the microkernel. What do you think they are planning on? I am of course not a Mac expert*, so I would be interested in your guess.
There’s talk of Apple doing all sorts of things. All the talk I’ve heard about Apple ditching the microkernel comes from non-Apple people who seem to have religious issues with microkernels. As far as I can tell, xnu is serving Apple just fine; and anybody who has an honest need for a monolithic kernel (or a true microkernel for that matter) is going to go somewhere else for their operating system needs. The vast majority of Apple’s market – desktop users and some pros – don’t know the difference and wouldn’t notice it even if they did. As far as I can tell, Apple doesn’t have a reason to care that they’re using a microkernel and not a monolithic one.
I think they’re planning on sticking with what they have; there are a bazillion other places where they could improve OS X’s performance more and with less effort than by replacing the kernel. Moving to a more fine-grained I/O locking scheme comes to mind; I’m sure that would do far more to help with your issues with database/webserver performance issues than switching to the FreeBSD kernel. Things have probably already changed quite a bit if you’re talking about experiences from a few years ago; it was only with Tiger that Apple finally introduced a lot of much-needed improvements in a kernel that was admittedly slow – for example, on OS X 10.3 and before, there was essentially only one huge “stop everything!” kernel lock, which is naturally absolutely murderous when you’re dealing a major PHP/MySQL web application. But that’s a stupid kernel locking scheme problem, not a microkernel problem.
Sad about Aperture, I liked the look of it, though never used it yet.
So it looks like Aperture had some quality problems, so they fired some of the engineers that worked on it. Fair enough.
What I want to know though is did they :
a) fire the management people involved in green-lighting a product noone was waiting for in the first place
b) fire the management types heading the team that ended up delivering this subpar (apparently) product.
I really want to believe they did, but I’m kinda cynical so I’m gonna day they probablty didn’t.
Apple Computer is guarding its server plans closely, leaving observers to wonder what will come of the product line.
And of course, Apple fans who publicly speculate on the company’s server plans will be sued. As in the past:
http://www.eff.org/Censorship/Apple_v_Does/faq.php
What you mean to say is if you publish trade secrets that were available only under NDA.
Think Secret is missreading the rumors.
The Aperture is a very innovating product, and professionals love it.
Apple have earlier done the same thing, removing the developer team from a product and replace it with others (more reliable devs). They do this when upper management is not very unsatisfied.
What you can expect is a Aperture 2.0 with a highly rewritten codebase.
Aperture is not a very different story than the other Pro apps.
Edited 2006-04-28 01:49
I would be really disappointed if Apple finally stops Aperture development.
New Apperture version was higly publicited during the last Steve Jobs keynote at Macworld and the product it is really good.
299 USD/299 EUR is a pretty good price for what you get, in my opinion.
Large Scale WebObjects deployment examples:
US Postal Service
AT&T Wireless before the Cingular days and before Apple buried WOF
Swiss Bank
Toyota
Dodge
Seattle Times
UPS
Department of Defense
The fact is most sites are Intranet based and you won’t be viewing them.