Sidenote: My apologies for the rambles herein. I think I ended up making a few points twice without cleaning up previous writing. It's a bit past my bed-time.
We can infer that it cost Apple a lot to run. -- a certain number of gigabytes or terabytes cost a lot more for someone like Apple to store (since they'll have to store it on fairly fast disks, that are reliable, presumably hooked up to a raid controller, on some kind of server system, and there'll have to be some kind of backup or restorable archive somewhere, for disaster recovery purposes -- elsewise, why have the information at all? They're not going to pop the archive up on a used G4 they bought on eBay and only expect one or two people per day to hit. Especially if Mac OS X 10.3 (and other older versions)(which could still be installed on a lot of network-connected machines) is set to automatically update. That would be getting potentially thousands of requests per day, worldwide.
We're talking about a fairly big system that needs to be fast with a high load.
Heck, what if Apple did have all of this stuff back in 2003 or so (and spent a small boatload on it) and when it came time to renew their hardware investment in this system (because maybe the hardware kicked the bucket, or was becoming too expensive to run), to provide updates for software a decade older or more, decided it wasn't worth whatever it was going to cost. (And given that Apple buys Sun/Dell/HP/whomever's hardware and runs linux/solaris on it for server duties, and that they'd probably want the support for the machine, they're going to be spending a lot.)
Not that Apple would ever publically talk about the statistics of such a service, but I bet the number of people hitting the 10.3 software update server who weren't fully patched has gone very very close to zero. Heck, when was the last time any of us here really did a clean install of 10.3?
My bet is that "most people' (who aren't computing hobbyists that buy old systems used and then reformat them with their period-appropriate operating systems) who still have these systems have been "fully patched" (as many patches as Apple will ever release) for years now.
sidenote: Stuff like this is part of why I've always considered it a good idea to, where possible, make and archive a backup of your full system in a working state on a regular basis. If something goes wrong with one of my machines, I rarely rebuild it and restore just the data -- I restore the whole machine and if my full system backup is old, I can then apply delta backups of the data.
I also build and archive full clean images of each system with all of the available updates/patches, the apps I'll generally want, and (where applicable) hardware drivers. This way if I get another system like it or really do need to start fresh (like the time I installed, uninstalled, changed versions, then did an upgrade on Visual Studio before Windows Update stopped working) -- doing so is quick and easy.
(although that is partly why I've started moving experiments into virtual machines.)
Support for 10.3 and everything before it ended a long time ago. It's arguably not really a mean-spirited move for Apple to actually end that support. That's what "support ended" means. Anyone doing production on that product either move off of it or prepare to do all of their own support. Apple having kept the stuff around so long was a nice gesture, not a mandate or promise.
The removal of a nice gesture isn't a mean gesture.
Even if it costs in the neighborhood of $10,000 to $20,000 (not counting the expense it may take to re-write or update the actual server component, since I'm pretty sure that won't run on modern linux/solaris/windows unmodified) -- that's $20,000 Apple could use somewhere else. ("drop in the ocean" aside) A very big component of the Steve Jobs Apple is that wherever legally and logistically possible, you don't look back. It's just a part of their corporate culture now, a part shared by many companies making consumer electronics.
We can infer that it cost Apple a lot to run. -- a certain number of gigabytes or terabytes cost a lot more for someone like Apple to store (since they'll have to store it on fairly fast disks, that are reliable, presumably hooked up to a raid controller, on some kind of server system, and there'll have to be some kind of backup or restorable archive somewhere, for disaster recovery purposes -- elsewise, why have the information at all? They're not going to pop the archive up on a used G4 they bought on eBay and only expect one or two people per day to hit. Especially if Mac OS X 10.3 (and other older versions)(which could still be installed on a lot of network-connected machines) is set to automatically update. That would be getting potentially thousands of requests per day, worldwide.
We're talking about a fairly big system that needs to be fast with a high load.
Heck, what if Apple did have all of this stuff back in 2003 or so (and spent a small boatload on it) and when it came time to renew their hardware investment in this system (because maybe the hardware kicked the bucket, or was becoming too expensive to run), to provide updates for software a decade older or more, decided it wasn't worth whatever it was going to cost. (And given that Apple buys Sun/Dell/HP/whomever's hardware and runs linux/solaris on it for server duties, and that they'd probably want the support for the machine, they're going to be spending a lot.)
Not that Apple would ever publically talk about the statistics of such a service, but I bet the number of people hitting the 10.3 software update server who weren't fully patched has gone very very close to zero. Heck, when was the last time any of us here really did a clean install of 10.3?
My bet is that "most people' (who aren't computing hobbyists that buy old systems used and then reformat them with their period-appropriate operating systems) who still have these systems have been "fully patched" (as many patches as Apple will ever release) for years now.
sidenote: Stuff like this is part of why I've always considered it a good idea to, where possible, make and archive a backup of your full system in a working state on a regular basis. If something goes wrong with one of my machines, I rarely rebuild it and restore just the data -- I restore the whole machine and if my full system backup is old, I can then apply delta backups of the data.
I also build and archive full clean images of each system with all of the available updates/patches, the apps I'll generally want, and (where applicable) hardware drivers. This way if I get another system like it or really do need to start fresh (like the time I installed, uninstalled, changed versions, then did an upgrade on Visual Studio before Windows Update stopped working) -- doing so is quick and easy.
(although that is partly why I've started moving experiments into virtual machines.)
Support for 10.3 and everything before it ended a long time ago. It's arguably not really a mean-spirited move for Apple to actually end that support. That's what "support ended" means. Anyone doing production on that product either move off of it or prepare to do all of their own support. Apple having kept the stuff around so long was a nice gesture, not a mandate or promise.
The removal of a nice gesture isn't a mean gesture.
Even if it costs in the neighborhood of $10,000 to $20,000 (not counting the expense it may take to re-write or update the actual server component, since I'm pretty sure that won't run on modern linux/solaris/windows unmodified) -- that's $20,000 Apple could use somewhere else. ("drop in the ocean" aside) A very big component of the Steve Jobs Apple is that wherever legally and logistically possible, you don't look back. It's just a part of their corporate culture now, a part shared by many companies making consumer electronics.