If distinguishing marks like serial numbers are blacked out, how could the leak be tracked down?
That is a very good point, covert tamper-proof seals could be in place. That said, there are still PowerPC -> Intel DTKs floating around in the wild. Compliance isn't going to be 100%.Well for starters, all the DTK's will eventually need to be returned to Apple, and if there are signs of tampering Apple isn't going to be happy.
That is a very good point, covert tamper-proof seals could be in place. That said, there are still PowerPC -> Intel DTKs floating around in the wild. Compliance isn't going to be 100%.
I guess if they strongly suspected someone, they might inspect a particular unit. For everyone else I doubt Apple will do much more than chuck the returned DTKs into an e-waste dumpster.
Don't have to. Apple has some large data centers. Rack them up and they'd have a decent compute cluster for some 'embarrassingly parallel" workloads. Run them for a year or two until there is a surplus of Apple Silicon Minis and incrementally swap them out of the cluster(s).
Similarly I suspect more than a few of these were in use at Apple for internal developer build aids before handing them back out (maybe a new case ). The design is probably been around for more than a year or so.
Given the state of macOS Server, I would be borderline shocked if Apple's data-centers actually ran on Macs.
(Sad but true )
I'm honestly surprised that no DTK teardowns have come out.
I think it's safe to say hundreds if not thousands of people have them, and surely someone would be willing to take the risk.
If distinguishing marks like serial numbers are blacked out, how could the leak be tracked down?
LOL they don't, they run on Azure and Google Cloud Platform.
Given the state of macOS Server, I would be borderline shocked if Apple's data-centers actually ran on Macs.
(Sad but true )
Not about the "outward" facing web services that Apple runs. Apple has relatively large and substantive inward facing servers too.
For example for chip development. Backdrop in this picture.
The Most Important Apple Executive You’ve Never Heard Of
www.bloomberg.com
Don't necessarily need MacOS Server for distributed computational modeling. ( Above are boards that have decent chanc e of having from FPGA attached to them, but indicative that Apple has internal needs that span past generic 1U - 2U generic rack models. That rack set up there is custom, not generic. )
Apple has more than few Apps for macOS and iOS that they distribute. All of that needs to be run through a QA testing. ( Continuous integration has side of effect of requirements for continuous testing. ) The DTK is a developer testing box. Which means is more than quite capable of doing development testing tasks. Got 10,000 QA tests want to run on an app... sent it to the "farm" to get worked over. For example, here is a set that Mac Stadium has.
Cloud-hosted Bare Metal Mac
MacStadium Bare Metal is dedicated, genuine Apple hardware hosted in MacStadium data centers.www.macstadium.com
About 23 per shelf. and 10 shelfs. So 230 Minis per. 10 of those racks on either side of an aisle isn't going to fill up a whole large data center but would soak up around 4,600 minis.
There is little to no evidence that Apple is spending too much money on QA. Yes, they need to have well composed tests to find more bugs earlier in the process, But with longer deeper regressions testing also will probably more hardware to execute the tests so the time to completion stays about the same. When these DTK's get to "too slow" to do that job then sure , eWaste them. But if they have decent performance then just chucking them is a weste.
[ There could be a hyper paranoid factor where Apple is afraid of trogan horse system coming back with "bad" firmware. Or something driving that as a security measure. ] Preformance wise though, doing iOS and iPadOS testing with this as a mule... it isn't going to lacking for more than a few.
If Apple wants disaster recovery, then one aisle in 2-3 data centers means they can "fail-over" if necessary with a 1-2 regional events.
Similarly depending upon how much code inspection and QA they run on App store submissions. That is a also a steady stream of stuff that doesn't require classic macOS server services. ( and closer to scaled developer build services. )
All of that has nothing to do with running iCloud drive services for 300M people. Or iMessage transaction brokerage . Or the other , "outward" , end-user Apple web services.
Interesting photo. Those boards look like they have the older style MagSafe connector on the providing power. I’d be curious to know the specs of those.
They do also have some of their own datacenters though, don't they? For instance: https://9to5mac.com/2015/02/23/apple-european-data-centers/
(I still doubt they contain Apple hardware, especially post Apple Silicon transition.)
Report: Apple designing its own servers to avoid snooping
Apple suspects that servers are intercepted and modified during shipping.arstechnica.com
Yes but they probably are operated by Google/Microsoft. I think they might be one of those on-premises solutions.
Hi,
Has anyone tried to take apart the DTK mac mini to see how it looks like inside ?
Thanks !
Considering how easy it is for Apple to trace any leak back to the original developer who leaked it, I doubt any developer would want to risk their developer account over it.I am really curious to see a teardown of the DTK too. or at least a sticker on a screw that prevents developers from unscrewing.. Seems we are getting none of that and in this age, i find it fascinating and amazing..
Even a teardowm won't tell us anything, as the DTK is not representative of the AS Macs. The DTK uses an old, off the shelf A12Z. I don't think you could have had 16GB of on SOC RAM on that device, so even if the RAM was off SoC, it says nothing about the AS Mac's SoC. Even if it were [pssible, Apple probably wouldn't want to create a new SoC just for the DTK.