See my previous comment, but it’s not “user unfriendly” as such. There are distinct advantages that they must believe outweigh the drawbacks. There are very few people who would both have an external bootable drive and have an internal SSD failure—and have a desktop machine, too, I think, since if you had a laptop, you’d be carting around an external drive if it the internal had failed.
Yes, it’s clear with an Intel Mac. I meant with M Series Macs onward and curious about having the option with them to easily replace your main drive with another, newer, faster, larger one.
The increasing confluence of iOS and Mac devices has many points where they meet, the approach to booting being yet another one.
You can absolutely run your M1 Mac from an external drive as the startup. The only reason you wouldn’t be able to is if the internal SSD failed. Which is unlikely, as I’ll keep repeating.
Onward, I don’t think it’ll change, because the M1 approach is just about managing secure policy, not at all preventing external drive use. The issue we’re really focused on is making bootable duplicates, and that (at least for now, probably worse in the future) will probably not improve. But that’s really distinct from a bootable external drive as your main drive.
Ah, it finally sinks in. Cheers.
In one of his posts on this topic, Howard makes the analogy that 1TR is essentially the same as the firmware on an Intel Mac – if the firmware dies/becomes corrupted, you can’t boot from an external disk. The difference with Apple Silicon Macs is that the firmware is no longer a separate chip, but a hidden partition(s) on the internal storage. Disk corruption on what you see as your ‘internal drive’ in MacOS is not going to cause problems for the 1TR ‘firmware’. As @glennf points out, if 1TR isn’t working you most likely have a significant hardware issue.
(There are obviously loads of differences in how the ‘firmware’ actually operates between the two platforms, but you can simplify for a high-level overview of whether it’s possible to boot from an external disk.)
And that issue is really about how difficult it is (maybe impossible in the future) to clone the system volume in Big Sur+ to your backup drive. You can ‘easily’ create a bootable duplicate by making a duplicate of your data volume and then installing Big Sur (or presumably future MacOS versions) onto the duplicate. The issue is that you then have to remember to boot into this duplicate and apply system updates whenever they’re released. There’s no way for drive cloning utilities to incrementally update the system volume on the duplicate by copying changed files from your boot drive.
By the way, great article Glenn – it brings this complex set of changes together in a single easy to read reference
analogy that 1TR is essentially the same as the firmware on an Intel Mac
If I’m installing Linux in the distant future, I’m less likely to accidentally format firmware, compared to essential partitions on an internal SSD.
As long as you stick to fuzzing with the container that holds the Data volume (disk0s2) you’ll be fine. You just want to make sure you keep your fingers from touching disk0s1.
I think somebody installing Linux on an ARM-based Mac is so far beyond mainstream that they can be expected to be knowledgable enough not to nuke the bootloader and “firmware” (well not so ‘firm’ now I guess).
But even if you accidentally do, it’s not the end of the world. Just get another Mac hooked up and use Configurator to clean up the mess and restore the initial setup.
Source of that beautiful schematic:
We think. Yes, the iBoot and iTR code appear as separate APFS containers on a single SSD, but teardowns have shown that there is a 64MB NOR flash chip alongside the multi-gigabit NAND flash chips that form the SSD.
We don’t yet know what that flash chip holds. If it’s the iBoot and 1TR containers, then we’re no worse off than what Intel Macs have. I’m hoping that someone will be able to figure this out and let the rest of us know.
In my view the ideal solution would be for Apple to make the SSD user-replaceable, maybe with some form of physical lock on the access door. That way you won’t end up with a brick if the SSD fails.
I totally get and agree with the security thing, but having a replaceable SSD would not compromise that as the data on the drive would be encrypted and there is no difference then between stealing the drive or stealing the whole system.
Either way you will need to have a backup if you are worried about the risk of theft, but at least with a replaceable SSD it could be replaced if it fails so that you aren’t left with junking the whole system and getting a new one if it is out of warranty.
IMHO this fear of failing SSD is exaggerated. Very few iPhones/iPads have suffered from it. Now with Macs I just don’t see why we should expect so much more. Sure we hold on to our Macs a bit longer and read/write patterns are different, but so different to expect 1-2 orders of magnitude difference in failure rates? Doubt it.
The real issue IMHO is that these days (ands this started well before M1) everything has to be configured right at initial purchase time. Later recognize you want more storage? You’ll need to buy a new Mac. On an iPhone that might be easier to handle because most people probably upgrade on a max 3-4 year schedule, but Macs are generally thought to have more life. Maybe no longer, at least in terms of initial owner. It will be interesting to see if this eventually puts some pressure on traditionally high Mac resale value.
My iPad (the only one I’ve ever owned) is as old as my MacBook Air – both from 2013. The iPad continues to work just fine, no failed SSD. It is a bit slow now and I’ll likely replace it soon, but the same is true of my MBA. I imagine there are a lot of people that hold onto iPads for a similar amount of time as their Macs and as you say, there aren’t loads of stories about failed SSDs.
Yeah, this is the real issue for me too. One of the main reasons I’ve not yet replaced my MBA with the new M1 ones is that I’m hoping the rumours of compact flash slots coming back to MacBooks are true. It’s useful for transferring photos from my Fujifilm camera, but the big reason I love the CF slot is that it provides a reasonably convenient way to add storage to a Mac laptop. Before I replaced the SSD in my MacBook Air with a larger one, I had a 256GB CF card I carried around all the time and used for a subsection of my files. (And of course it was included in my backup strategy!)
It’s important to realize that what we call an “SSD” on an M1-based Mac (and Apple laptops in general, for quite some time) is really just one or more chips soldered onto the motherboard. There is nothing anyone would recognize (or that any other company sells as) as an SSD that could be removed and replaced.
Yes, pretty much my point. It would in my view be better if this were not the case.
What no-one seems to know (as far as I can find out) is what the interface is between the SOC and the “SSD” (or whatever one calls it!). Assuming it is not some form of direct addressing from the memory management circuitry (which maybe it could be) it is likely that there is a bus which could be implemented through a connector rather than directly soldered connections. Yes, it might be very slightly less reliable, and cost very slightly more to manufacture, but at stake is the reputation of M1 Macs.
In time we will know, but if there is any likelihood at all that the “SSDs” will fail before people feel they should do (7-8 years maybe?), resulting in bricked systems, Apple Silicon based Macs are going to get a drubbing from the press and anyone else who is out to get Apple (of course no-one out there wants to do that, do they?).
Apple must be incredibly confident that these SSDs will last for the life of all the other components. Let’s hope that confidence is not misplaced. I am personally apprehensive about getting an M1 Mac at present purely because of this issue.
This isn’t an M1 Mac issue, so I’m not sure how this will impact the reputation of them as opposed to Macs in general. The ‘SSD’ on MacBooks has been some chips soldered to the logic board since around 2016 I believe, so going on for five years now. So far I’ve not heard that failure of these storage chips has been an issue.
Also, in your list of benefits, you left off one that I think is key to why Apple has gone this direction: size/space. Soldering the storage chips directly to the logic board saves a lot of space versus a connector and supporting card. And not just the reduction in actual space, but there’s an increase in flexibility as to where they go and the overall layout of the machine. I think if Apple could offer replaceable SSDs and not have to sacrifice any space or layout flexibility, they would have continued to do so (ditto for the battery).
Apple has no motivation to do this and it’s clear from the last several years, it’s not a direction that they feel has any value in taking, or they would have used the M1 as the opportunity to do so with a clean slate.
I realize it’s frustrating to lack an upgrade path, as I have certainly felt that with my last two non-upgradable laptops, when I finally left the “swap a hard drive or SSD” pathway. But Apple is almost always looking for the intersection of manufacturing complexity reduction, material reduction, price, and mass market.
Relatively few people need to update the storage on their laptop drives in the current era of 500GB and 1TB drives predominating; Apple wants to push them into upgrading a laptop. I don’t know this as a fact—but it’s a supposition given Apple’s high ratings among customers in published surveys and the fact that it addresses product needs by changing out products.
If it were truly a thing Apple needed to do, they could have a side module you could pop out and stick a new SSD module into, I’m sure.
On the desktop and via Thunderbolt 3 there are tons of opportunities for drive expansion, so it’s as if Apple is channeling that desire into that particular path.
Fewer connections results in fewer failures, too. The more you have to package something and plug it together, the more likely some portion may be a weak link.
The difference now is that the system is dead if the SSD fails. That wasn’t the case previously, and that’s the difference for me. You’re all correct with everything you say, but the stakes are higher now.
And, as signal voltages are made smaller, connector corrosion becomes a primary failure mechanism.
I’ve read this article a few times. I’m not seeing how this (to be specific the 1TR partition) is making the Mac more secure.