On Dec 26, 2011, at 4:32 PM, Michael_google gmail_Gersten wrote:
> 
> As for the immediate "Lock-in":
> 1. Apple has decided that my iOS device, that I purchased, that I own,
> can only run software installed via the AppStore. It's my device, but
> they want to control what I do. (Assuming I buy one, anyways; haven't
> yet).

Decided in advance of sale. Known in advance of purchase. No change. If you 
don't like the EULA, and the residual claim to ownership of a portion of the 
property you purchased, don't buy one. Or buy one 2nd hand so you aren't 
agreeing to the EULA, and then jailbreak it. I have a much bigger complaint 
about them becoming international patent trolls, than their EULA.

It's a different matter when things change after the fact, like they have been 
on desktop. But then, what promises is Apple making? Slim to none. It's valid 
to question, but you will get no answers, at what threshold desktop is 
relegated to the same dustbin as enterprise products and various other 
technologies Apple has attempted and then dumped. *shrug*



> 2. IBM got into trouble for requiring that the computer and OS be sold
> as a unit. They had to split that up. Microsoft got into trouble for
> requiring that the computer maker had to purchase and sell a license
> for every machine they sold. Despite a victory in the courts, the
> marketplace made it moot for consumers who still cannot (in reality)
> get a machine without a Microsoft license without a special order. The
> justice department did not follow up to make that consumer victory
> real. Now Apple is doing the same thing -- requiring an OS purchase
> even if you want to run Linux -- and getting away with it.

Not the same thing. On multiple levels. Apple sells a specific product: 
hardware and software. You only get Apple hardware from Apple and you only get 
Apple software from Apple. When you buy from Dell, it's obvious to a house 
plant that Dell gets Windows from Microsoft for a price and there is a line 
item cost for Windows. So it stands to reason consumers can say they decline 
purchasing that particular line item. There is no such line item for Apple. 
It's included, it has no cost, you get it.

Also, it has been this way since the beginning of Macintosh. You are not buying 
hardware. You are not buying an operating system. You are buying an experience 
that combines the two, which are inseparable. Just like iPhones and iPads and 
iPods. Inseparable combination of technologies that form an experienced, sold 
as a product. Apple is the hallmark company that denies its customers have any 
right to operating system freedom, to boot whatever they want. Not news.

> 
> Apple's OS has been shown to run on umpteen different makes and
> models, but is only permitted to run on about 6. All from one company.
> Any attempt to do otherwise is met with a lawsuit that starts with the
> assumption, "We have the right to restrict how you use something that
> we sold you for your use".

Nope. The EULA says you have a license, not right, to use Mac OS X on Apple 
hardware. You either agreed with the EULA or not. 

I'm not making a case that Apple is legally or ethically obligated to do 
something they never have before. I'm saying, they should eat their own dog 
food, and think different, and offer a new restrictive license to run Snow 
Leopard Server in VM on non-Apple hardware in order to placate those who made 
major investments in Apple server hardware and now no longer have a hardware 
upgrade path at all. It would be a self limiting license. It would make sense 
for them to put in an expiration date - give it 4 years and the license 
expires. But they are under no moral or legal obligation to do this. It's just 
good business rather than quitting servers cold turkey. But hey, they made no 
promises so we all should have known better in the first place.

> You can't say that an automaker requires you to use their motor oil,
> or their service stations, to maintain your car.

Old arguments, not the same thing. An operating system is not a consumable.

> You can't say that a printer maker requires you to use their inks in
> their printers.

Tell that to Epson. They get around it with encrypted programmed ink cartridges.

> 4. If I want to use an Apple software technology, I'm required to use
> an Apple hardware technology.

Known in advance. 

> 5. Even with the Apple hardware, and Apple software, there can be a
> driver bug; Apple won't fix it. (ATI video drivers on older hardware.)
> Customers have no recourse.

Actually customers have no recourse for bugs from any company insofar as I'm 
aware. All EULAs say they have no liability for any software or hardware 
defect. Apple isn't unique.

> There is a difference between "Tightly integrated hardware and
> software", and "never mind that you can get it to run elsewhere, we
> won't let you; never mind that you want to do X, we won't let you;
> never mind that you might want to run Linux on our hardware, you still
> have to buy an OS that you don't want to use".

I'm not going to accept the argument that you buy Apple hardware for the 
purposes of exclusively running some other operating system. I've already 
relayed the crappy firmware situation, there is no good reason why anyone would 
buy Apple hardware with the express intent to remove OS X (not use it at all) 
and instead use linux. Just don't buy it. It's a bad business decision to try 
and run foreign OS's on Apple hardware. Red Hat has a huge laundry list of 
hardware that they've certified for RHEL, which is sufficient to run most any 
linux including Fedora, Ubuntu, Mint - etc.


> 
>>> Where I think Apple is _FAIL_ is simple: Why can't I run older
>>> software in a VM on current Apple hardware?
>> 
>> I'm not sure I understand. Are you saying with new hardware that requires 
>> 10.7, that you can't virtualize 10.6 server on that hardware? That would be 
>> ... slightly interesting to see what snotty thing I'd come up with if that 
>> were true.
>> 
>> My complaint is being unable to legally run 10.6 server in VM on non-Apple 
>> hardware. As in, a real server.
> 
> Actually, when the new 10.7 airs came out, people were saying that
> 10.6 would not run at all, even in a VM.
> 
> Now, my complaint is being unable to legally run 10.6 user in VM on
> *APPLE* hardware. As in, a Rosetta compatible upgrade. As in, a
> work-around for the limitations of 10.7's new "features".

You can, it just has to be 10.6 server. That EULA allows it to be run in VM.

You're correct, it would be nice if the client EULA were revised to allow 10.6 
to be run in VM, until 2015. But I will bet dollars to donuts Apple knows 
client will run fine in a VM, and they don't care. You can do it. Yes it's a 
violation of the EULA so a big business can't do that and risk liability. But 
while Apple is sue happy in many other areas, I don't think this is one of 
them. If you need to do it, you could do it, violating your EULA.

Why won't Apple change the EULA? I will bet licensed technologies. I'm sure 
there is an expected life span for an operating system, just like hard drive 
manufacturers put one on hard drives. And those licenses expire so Apple can't 
produce a modified EULA that effectively allows things like Rosetta to run for 
eternity in VMs. The company Apple negotiated licensing Rosetta no longer 
exists, owned by IBM now. So maybe that's got something to do with it. They 
just don't want to renegotiate the licensing terms. No incentive.

> Because, oddly, Apple will sue you and tell you that you should have
> made your tablet non-rectangular, non-clean looking, and a host of
> other stupidities. See the current (Samsung, I think?) legal fight
> over iOS interface / iPad design patents.

I don't think Apple can sue a company over the smoothness of their hardware, 
nor do I think they've tried. They have tried suing based on similarity.



> 
>> Competition law is probably out of scope because well before a legitimate 
>> conversation on competition law occurs, one has to establish a common frame 
>> of reference on vastly more basic things like property, and whether free 
>> markets can even exist ... And besides that, since when does the U.S. 
>> government exhibit any consistency whatsoever on antitrust?
> 
> Consistency: Exactly my issue.
> Free market existence: While a perfect free market, just like a
> perfect democracy, cannot exist, you can get much, much closer to it
> than we have today. We even have laws that are supposed to require
> competition and free markets. They are not enforced.

This is the nature of the American political system. The ability to execute 
laws independent of their existence is built into the system. It's as clear as 
mud we could do much better. We could also have a regression and do much worse.


> 
>> Ehh. Linux distros have their own issues outside of GUI that makes it pretty 
>> rough on normal people.
> Well, are you still required/expected to recompile the kernel to match
> your hardware configuration, or does it finally adjust to your
> hardware at bootup / configuration time?

You can build kernels that include kernel modules for quite a wide assortment 
of hardware, if they are required to be compiled into the kernel. Otherwise the 
initramfs is used for most of this. This is why it's not unusual to see a 4MB 
kernel on a Live CD, but see a 150MB initramfs that has a bunch of hardware 
drivers on it. This is also how it's possible for distributions to issue one 
distro specific kernel, and upon installation of the updated kernel, a new 
initramfs is produced, which is fairly computer specific.

Sever, security and performance enthusiasts will still compile their own kernel 
with specific features enabled and disabled. It's quite common.



> Does it auto-detect hardware
> plugged in, and setup drivers automatically, or do you have to run a
> config program to scan the buses and reconfigure drivers?

It does auto-detect hardware when plugged in, and loads appropriate drivers, in 
most instances. 


> I'm assuming
> you're not required to pass boot args to get the kernel to run
> anymore, right?

This is still required as is it for XNU. If you look at your NVRAM there are 
things passed to the bootloader and kernel.

For linux, it's necessary to pass at least one basic kernel argument which is 
the location of the root directory. The bootloader will load the kernel and 
initramfs into memory, but neither of them  have any idea what to do after 
that. They are told with a kernel parameter root=/dev/xxxxxx  where that can be 
an actual device, pxe device, lvm, dm-crypt/LUKS mapper, RAID, iSCSI initiated 
connection, or various other locations for loading the rest of the operating 
system.

If you don't have a custom kernel, but have some hardware that otherwise would 
make it necessary, there are kernel parameters that can be passed to the kernel 
instead of compiling a custom kernel to get the behavior you want.

It's also common to use kernel parameters to send a message to dracut to boost 
the performance of boot by telling the kernel what NOT to scan for on startup. 
If you don't use LVM, or RAID, or LUKS, you can tell it this and then it won't 
scan for all possible locations for root.

So they are still used.



> 
> Just a quickie: How are people treating Network-attached storage (NAS)
> as different from Storage area networks (SAN)? What do you see as the
> difference?

NAS uses a network file sharing protocol like AFP, NFS, SMB, which runs usually 
on Ethernet over TCP/IP.

SAN uses a clustered file system like Lustre, GFS2, or Blue Whale. Client 
workstations have direct or indirect access to the storage as a block level 
device. Data blocks are carried over fibre channel, or iSCSI, whereas the file 
system metadata is carried over a dedicated ethernet network.  It appears and 
acts as a locally attached disk, but all clients can simultaneously share the 
same blocks. The clustered file system deals with lockouts.  Distributed 
varieties will also have fallover so the data remains available even if some 
nodes fail. With a NAS, if the NAS server dies - well you're SOL until you get 
it back up and running again.


> Looking over "SAN vs. NAS" at Google, the difference seems to be that
> a SAN is a single network attached controller that talks to many
> drives, while a NAS is many networked attached controllers each
> attached to a single drive. Is that it?

A SAN will have at least one metadata controller, preferably two in any 
substantial setup in case one goes down. XSan recommends two a primary and 
standby metadata controller. Only fibre channel for data block access. And 
either dedicated or shared ethernet for the file system metadata connection.

GFS2 implementations could have dozens of them, and many petabytes of storage 
in multiple physical locations, all of which is available all of the time even 
if metadata controllers or LUNs fail.


> But speed wise, if they are both over 10 GB ethernet, aren't they both the 
> same?

Different topologies, protocols, purposes, implementations, and failure 
behavior. 


Chris Murphy
_______________________________________________
MacOSX-admin mailing list
[email protected]
http://www.omnigroup.com/mailman/listinfo/macosx-admin

Reply via email to