MacInTouch Amazon link...

storage systems

Channels
Products, Questions
How high a percentage of that 4TB Time Machine drive is being used? If you increase the provisioned areas from 2 TB to 3 or 4 TB, is the Time Machine drive now too small? I think the cap for 2.5" hard drives is 5 TB . For the 5-8TB range for a Time Machine drive, switching to 3.5" hard drives might be better. A 3.5" 6TB Iron Wolf drive and 2.5" 5TB Barracuda cost approximately the same.
Currently the Time Machine drive is backing up the 2TB (usable) RAID4, 256GB internal SSD in the iMac, and (via Server.app) the 500GB SSD in a MacBook Pro, for a total of 2.75 TB of nominal capacity and probably around 2 TB of actual data. It isn't big enough for a long backup history, but it's plenty sufficient for my needs.

And while 5TB is the largest 2.5" consumer hard drive currently available, the 4TB I have is already a replacement for a 3.5" drive I was previously using for backup. Cost has nothing to do with it--the 3.5" version was just unnecessarily noisy, even in a fanless enclosure on a styrene pad. The 2.5" drive is much quieter, and I'd rather just stick to smaller, lower-power storage than try to build a soundproof enclosure or something.

Were I to go to a RAID0 + Time Machine option, I would probably get a second 4TB or 5TB hard drive for the fourth bay to use exclusively for backup of the SSD RAID, and delegate the drive I already have to backups of the internal SSDs and/or extra backup or scratch files or something. With the 3xSSD RAID4/5 option I'd be forced to get a 5TB hard drive for the Time Machine drive if and when it got close to full, but that might be several years from now, by which point I could do something else.

In any case, in the next two years the issue isn't backup capacity - I have enough of that - it's how to get the online storage to SSD without costing a fortune.

I think one of the root causes missed in your list of options is the issue of the drive bay that the Time Machine sucks up. That's part of the capacity problem. If you striped RAID 4 over 3 bays with 1TB drives, you'd have 3 TB of usable space.
You may have mis-typed, but three 1TB drives striped as RAID4 is exactly what I have already, and it only provides 2 TB of usable space, not 3 TB, since the third drive is for parity. I only get 3 TB if I fill all four bays with drives and move the backup drive to an external enclosure.

Which of course I could do - I could get four 1TB SSDs for 3TB of usable space. That's a 50% capacity increase and only requires 33% overprovisioning rather than 50% to maintain capacity. But I'm quite reluctant to spend nearly $800 for a 50% increase in storage space (and some speed) when I'll be back to having to replace the works in probably 2-4 years.

If I settled for RAID0 + Time Machine, I get double the storage for around the same cost, either have an extra tier of backup or an extra bay available, don't need another external, and don't have to think about refreshing storage for probably 5 years, when I'd be thinking about new drives anyway.

A NAS box as a Time Machine target could be moved away from being highly local, if noise is an issue.
I'd seriously considered this option (added advantage: physical separation of backup and online data), although the unnecessary waste of power of an always-on NAS box annoys me, ones with good Time Machine support tend to be kind of pricey, and I've previously had very poor experience with this configuration.

A workflow change, but another relatively low option would be to use two 2TB SSDs in RAID 1 for critical near-line stuff and to devote one bay to "intermediate candidates for a 3.5-inch box" with a 2TB drive and then a 5TB Time Machine drive that backs up both of those two volumes.
For practical purposes I already have exactly this, except the nearline storage is the 15TB (usable) Thunderbay unit that I don't normally have powered on and has plenty of free space on it. Adding a third tier isn't anywhere worth the mental load of having to pay attention to what's stored where; it's enough as-is to say "Either it's always there or I need to flip a switch to get to it".
 


Ric Ford

MacInTouch
Samsung's new Thunderbolt 3 Portable SSD X5 is not compatible with older Thunderbolt 1 or 2 Macs, as neither Apple's Thunderbolt 3-to-Thunderbolt 2 adapter nor third-party alternatives work. Apple's adapter does not supply bus power, which is required by the SSD, and alternative adapters work in the opposite direction only (connecting a Thunderbolt 3 computer to a Thunderbolt 1 or 2 peripheral).
We tried the X5 on a Dell gaming laptop, which has Thunderbolt 3 port, and it showed up fine and offered various software options and status updates. However, it also keep disconnecting, though we had the included Thunderbolt 3 cable connected firmly. Thunderbolt disconnects have been an ongoing theme with a variety of products and Mac systems. It's not clear if this is some kind of Thunderbolt driver/firmware issue with the Dell or a problem with the x5 itself.
 


Thunderbolt disconnects have been an ongoing theme with a variety of products and Mac systems.
I've certainly read enough anecdotes to know that this is true, but it's also worth keeping in mind that there are plenty of people who use Thunderbolt peripherals without disconnect issues.

I have a 2014 iMac at home that is put to sleep multiple times daily with a RAID array of always-mounted data drives in a Thunderbay Mini that has not had a dropout in around 2 years of use. At work we have an always-on 2012 Mini fileserver with a Thunderbay 4 in similar config that hasn't had an issue in a little over a year of constant use, a Mac Pro Can with another Thunderbay that hasn't had dropouts since it was purchased in 2013, and an iMac that has been using a bus-powered Monster Thunderbolt SSD as a boot drive for around 2 years without issue.

Admittedly, those are all Thunderbolt 2 devices, 3 of 4 are OWC, and it's, of course, only anecdotal, but disconnects certainly aren't a given, or a fundamental problem with the bus, OS, or Mac hardware -- something causes them.
 


We tried the X5 on a Dell gaming laptop, which has Thunderbolt 3 port, ... However, it also keep disconnecting, though we had the included Thunderbolt 3 cable connected firmly. ...
Is this a recent Dell laptop or one from the relatively very early Thunderbolt 3 era? One of the issues with Thunderbolt 3 has been merging with USB on new (expanded) power regulation implementations and standards. Early on, there was a single TI power regulation chip used in most Thunderbolt v3 implementations that went through several revisions. I think that was a contributing reason to why Apple didn't arrive in the very first wave of Thunderbolt 3 implementors.
It's not clear if this is some kind of Thunderbolt driver/firmware issue with the Dell or a problem with the x5 itself.
The more expansive the ecosystem of products, the more likely to get interaction quirks.
 


Ric Ford

MacInTouch
Is this a recent Dell laptop or one from the relatively very early Thunderbolt 3 era?
It's this one, purchased in March, which I like a lot, as it's completely accessible and upgradable (one-screw access), has a nice screen and keyboard (and a keypad), plenty of ports (including Thunderbolt 3), has both M.2 and SATA bays, ample cooling, etc.

Dell Laptop - 7th Gen Intel Core i5, GTX 1060 6GB Graphics, 8GB Memory, 128GB SSD + 1TB hard disk drive, 15.6"

The owner uses it for video editing (1080p) in Premiere Pro and says that it's as fast as a new 27" iMac for that purpose when connected to a/c power (but is slower on battery). His media goes on the hard drive, with the system on the SSD.
 


A NAS box as a Time Machine target could be moved away from being highly local, if noise is an issue.
I have a Drobo 5N2 NAS that I set up to use as a Time Machine backup device, as well as to hold rather large Photos and iTunes libraries. I installed five 4TB WD Red drives, which, with double-redundancy in the RAID, provides for a bit more than 10TB. (I could have a fair amount more if I configured it to work with just one failed drive at a time.)

The Drobo software allows for creating a Time Machine volume, of fixed capacity, on the drive, and that seems to be working perfectly. Currently, it's backing up two machines as a secondary Time Machine backup device. (Each machine alternates backups between a 3TB Time Capsule and the 3TB Time Machine volume I set up on the Drobo.)

I also set up a separate share on the Drobo to use for Photos and iTunes. I mounted the drive, using the Finder, on the machine from which I want to manage these libraries, and the drive kept disconnecting at random times. After some back and forth with Drobo, I installed the "Drobo Dashboard" on that machine and mounted the volume using that. Since doing so, it's not disconnected, and it does auto-mount on log-in, which is what I wanted, anyway. (Neither of the machines using Time Machine has seemed to have a problem with that volume, but the mounting is handled by Time Machine, and it does not have to remain connected beyond a given backup.)

Once I managed to get a stable connection going, I copied my Photos Library to that volume. I opened Photos while holding down the Option key and selected that Library. It took forever to open, but eventually, it did. Once it was open, it seemed to reopen easily enough. I tried to set that library to be my System Library, so I could turn on iCloud Photo Library, and I received a notice stating that the volume must be formatted as Mac OS Extended or as APFS.

When I contacted Drobo, they said that since it's a NAS, the share cannot be reformatted.

So, I've created a disk image file on the share, formatted the image as Mac OS Extended, copied my libraries there, and I'm testing that out, now. (So far, Photos seems to be happy with it, but it's just started syncing.) I've not tested iTunes, yet.

The Drobo is in a large closet, so I've not heard anything from it, but even from up close, it's not terribly loud; just some low fan noise.

I will have to see if the disk image will mount automatically. (I've added it to my startup items on that machine, so I'm hopeful; I just don't know how fast that might happen upon login.)

All in all, the Drobo was easy enough to setup, and if it provides the redundancy and reliability that's promised, it should be perfect for my needs. It was not a small investment, but far less than anything of this capacity and capability was a few years ago.
 


Ric Ford

MacInTouch
Admittedly, those are all Thunderbolt 2 devices, 3 of 4 are OWC, and it's, of course, only anecdotal, but disconnects certainly aren't a given, or a fundamental problem with the bus, OS, or Mac hardware -- something causes them.
Maybe it's cellphones...
Amazon reviews said:
OWC 12 Port Thunderbolt 2 Dock...
....
2) This is the big one. The dock will just randomly disconnect while I am using it. The cables are not loose or moving. My second monitor will start flickering, then go black and then after a few seconds everything will come back. This usually happens once a day, a few times in a row and then stops. If it happened more frequently I would probably throw this thing away. Occasionally, to fix this, I have to unplug everything and wait a little bit, not good!

Anyway that's all. This thing definitely does what it says, it's just not that reliable.

Edit:
So I seem to have discovered the cause of issue #2. By chance the other day I noticed that the dock freaked out as I described above, seconds after I put my cell phone next to it. I am an Android developer and frequently have my phone right next to where the dock is placed on my desk. I assumed it was a coincidence and but decided to do some testing anyway. I found that the docks usually freaked out within 15 seconds of me putting my phone near it. Since then I have been keeping my phone on the other side of my desk and the issue has only happened once or twice in the entire week. Still disappointed with this thing but at least I have discovered a work around. Sharing it for any one else who is experiencing this issue.

Steve Overby 4 months ago.
I've been having issue #2 lately after using my dock for about a year. I just put my phone next to it and it did the freak out. So I wanted to get your take some 2.5 years later; does this still seem to be the problem?
 


I have been tracking Thunderbolt "disk eject" and bus reset issues for several years. Here is a quick summary:

This problem was first identified in 2016. There was a clear identifier in the system log, of the PCI bus resetting, which ejected all Thunderbolt devices from that bus. After a long struggle, we got Apple to address this issue in macOS 10.12.2.

However, while Apple eliminated probably 80% of the incidents of disks ejecting, this problem remains, but is no longer leaving any traces in logs, so it is much harder to isolate the root cause.

It is still relatively rare, but in a new SoftRAID update, we’re adding a tool for anonymous data collection that will allow us to track this issue statistically. This means, for the first time, that we will be able to demonstrate what % of computers with Thunderbolt devices have this issue.

Any device or computer can be the "culprit". Generally, however, this problem becomes more common the more devices you have. The most common trigger seems to be second or third Thunderbolt monitors.

I have had multiple users report that disk ejects start/return with OS X updates! So this is not purely a hardware issue.

I have had reports of disk ejects on nearly every brand of enclosure. It happens sometimes with Thunderbolt docks. Often, a user with one Thunderbolt enclosure working fine starts seeing this problem when adding more devices.

It happens with Thunderbolt 2 and 3 both.

We have Apple's attention on one specific scenario, which I would appreciate assistance getting more data points on, which is Thunderbolt monitors. (Note: If you have an LG monitor, make sure it has been serviced, or upgraded, as certain models are known to trigger disk eject events.)

If you have two or more monitors and experience disk eject events, try this simple test:
Disconnect the second monitor. Do you have less events? If you can demonstrate significantly less disk eject events, then we can forward your data to Apple for further investigation by their team. (There is specific data we first need to collect in order to get this to Apple engineering.)

For some reason, Final Cut Pro editing seems to increase the frequency of disk eject events.

What can you do?
Apple recommends two things:
1. Clean all connections with 90% isopropyl alcohol (which is difficult to do with Thunderbolt 3, as the USB C connectors are so small).​
2. Reset NVRAM.​

We notice also that disk eject events are more likely to occur after waking from sleep, or after intensive I/O to the disks (like large file transfers), then when the disks are idle for a while, a disk eject may occur. (So setting the system not to sleep can help.)

My experience is that these measures help in only a handful of circumstances, but are helpful where the problem starts immediately after an OS X update (or security update/kernel driver install).

Replacing enclosures may work, but generally does not. There is an interaction of some kind that causes this, and this problem does not seem to be tied to a specific enclosure. An example of this is that I have arranged for several users with serious problems to exchange their OWC enclosures (Thunderbays), for new ones. OWC was not able to reproduce the problem in their test lab, and the problem remained with the user. So this can be very frustrating on all ends.

I invite anyone experiencing disk eject issues to email support at softraid. You do not need to be a SoftRAID user for us to try to help. We all want the same thing, which is for Apple to acknowledge and fix this incredibly annoying problem.

Note: Placing a cell phone next to a dock is a known problem with most docks. The signal frequencies are close enough to cause interference.
 


Good morning, I am seeking a 5-port eSATA to SATA port multiplier for use inside an external enclosure. I've tried several, and so far, cannot find one that will work with macOS Sierra on a 15" MacBook Pro (13,3). My external eSATA host controller is a Thunderbolt 3 dock from CalDigit, which works great with Open Digital Mobius 5 external hard drive enclosures (eSATA, USB 3, etc.). The Mobius drives show up in JBOD mode as well as RAID arrays, depending on how the jumpers in the back of the enclosure were configured. Here is what I have tried:

1) Coolgear Sil 4726: This controller is pretty ancient. The sellers claim that management software to configure it is available from the OEM (now Lattice Semi), when in fact, it's not. Without being able to configure this controller for JBOD use, it won't mount any drives. If you troll the web, you can find OS X versions of the manager software. Unfortunately, it's PowerPC code and hence will only run on Mac OS X Snow Leopard and before. There is a Windows utility, but I don't have a Windows computer around with eSATA. So, no go.

2) JMB321-based generic controller: This controller is really interesting: It successfully crashes my computer every time I plug it in. I replicated this behavior with a Kanex eSATA external host, as well as the Thunderbolt 3 dock from Caldigit. In both cases, the computer screen will go blank and the machine will restart. No kernel panic, nothing. Just blank screen and reboot.

I'd be delighted if the controller also offered USB 3 as well as eSATA external interfaces. The cherry on top would be a controller that uses one of those SCSI-1 brackets to allow easy mounting to a multi-bay DVD enclosure (but that's not necessary). However, this has to be a standalone controller, i.e. it should not require being plugged into a PCIe bus to operate.
 


Good morning, I am seeking a 5-port eSATA to SATA port multiplier for use inside an external enclosure....
After some additional searching, I came across Addonics, which makes a controller (HPM-XU) which features both USB 3 as well as eSATA external interfaces, configured with DIP switches, and mounted on a SCSI-1-sized bracket. It'll be perfect, if it works!

The HPM-XU costs $69 and can also be configured / upgraded, etc. via the JMicron Raid Manager software hosted at Addonics. They show a firmware there for a USB sleep issue and I have used this manager before to configure/upgrade Oyen Digital Mobius 5 enclosures.

Meanwhile, I am requesting an RMA for the Sil4726 controller. The generic one isn't worth the trouble.
 


Good morning, I am seeking a 5-port eSATA to SATA port multiplier for use inside an external enclosure. I've tried several, and so far, cannot find one that will work with macOS Sierra on a 15" MacBook Pro (13,3). My external eSATA host controller is a Thunderbolt 3 dock from CalDigit, which works great with Open Digital Mobius 5 external hard drive enclosures (eSATA, USB 3, etc.). The Mobius drives show up in JBOD mode as well as RAID arrays, depending on how the jumpers in the back of the enclosure were configured. Here is what I have tried:
1) Coolgear Sil 4726: This controller is pretty ancient. The sellers claim that management software to configure it is available from the OEM (now Lattice Semi), when in fact, it's not. Without being able to configure this controller for JBOD use, it won't mount any drives. If you troll the web, you can find OS X versions of the manager software. Unfortunately, it's PowerPC code and hence will only run on Mac OS X Snow Leopard and before. There is a Windows utility, but I don't have a Windows computer around with eSATA. So, no go.
2) JMB321-based generic controller: This controller is really interesting: It successfully crashes my computer every time I plug it in. I replicated this behavior with a Kanex eSATA external host, as well as the Thunderbolt 3 dock from Caldigit. In both cases, the computer screen will go blank and the machine will restart. No kernel panic, nothing. Just blank screen and reboot.
I'd be delighted if the controller also offered USB 3 as well as eSATA external interfaces. The cherry on top would be a controller that uses one of those SCSI-1 brackets to allow easy mounting to a multi-bay DVD enclosure (but that's not necessary). However, this has to be a standalone controller, i.e. it should not require being plugged into a PCIe bus to operate.
Port Multiplier... OS X driver... Thunderbolt 3 aware...That's serious old school meets new school. I bet there is a "newer" chain to make that work or... build a server... 10-Gig FreeNAS box with a 10-Gig card and put a 10-Gig connector on the laptop. (Or connect via simple GigE.)
 


After some additional searching, I came across Addonics, which makes a controller (HPM-XU) which features both USB 3 as well as eSATA external interfaces, configured with DIP switches, and mounted on a SCSI-1-sized bracket. It'll be perfect, if it works!

The HPM-XU costs $69 and can also be configured / upgraded, etc. via the JMicron Raid Manager software hosted at Addonics. They show a firmware there for a USB sleep issue and I have used this manager before to configure/upgrade Oyen Digital Mobius 5 enclosures.

Meanwhile, I am requesting an RMA for the Sil4726 controller. The generic one isn't worth the trouble.
I don't see the PCIe card that goes into the Thunderbolt box. This Addonics is just for the external box. The PCIe card for the Mac is the tricky part.
 


I don't see the PCIe card that goes into the Thunderbolt box. This Addonics is just for the external box. The PCIe card for the Mac is the tricky part.
... The eSATA host interface is built into the Thunderbolt Station 3 dock from Caldigit. The dock features two eSATA ports; both are port-multiplier enabled. The external enclosure is the one that needs the port-multiplier controller.

FWIW, my enclosure is populated with 4 blu-ray drives and one hot-swap 3.5" bay. I use those blu-ray drives to encode my music collection; the 3.5" hot-swap bay is to burn in new hard drives or zero out failing ones. FreeNAS is unlikely to help me for this application (it's great software though).

If you're looking for good external PCIe enclosures for connecting Thunderbolt-aware gear to a Mac, I had a very good experience with the Thunderbolt 2-based Helios PCIe enclosure from OWC (a slightly-rebranded private label production from Akitio) and currently enjoy the Highpoint Rocketstor 6661 for my Myricom 10GBe network adapter. Curiously, the current Helios 3 from OWC is being sold for less money than either the Rocketstor or the original Helios enclosure. Go figure.

Lastly, I disagree with characterizing SATA/eSATA as somehow over the hill. Unlike USB and other connection methods, eSATA provides a very inexpensive, fast, reliable interface that also allows my computer to detect SMART error issues as they come up.
 


Good morning, I am seeking a 5-port eSATA to SATA port multiplier for use inside an external enclosure. I've tried several, and so far, cannot find one that will work with macOS Sierra on a 15" MacBook Pro (13,3). My external eSATA host controller is a Thunderbolt 3 dock from CalDigit, which works great with Open Digital Mobius 5 external hard drive enclosures (eSATA, USB 3, etc.). The Mobius drives show up in JBOD mode as well as RAID arrays, depending on how the jumpers in the back of the enclosure were configured. Here is what I have tried:

1) Coolgear Sil 4726: This controller is pretty ancient. The sellers claim that management software to configure it is available from the OEM (now Lattice Semi), when in fact, it's not. Without being able to configure this controller for JBOD use, it won't mount any drives. If you troll the web, you can find OS X versions of the manager software. Unfortunately, it's PowerPC code and hence will only run on Mac OS X Snow Leopard and before. There is a Windows utility, but I don't have a Windows computer around with eSATA. So, no go.

2) JMB321-based generic controller: This controller is really interesting: It successfully crashes my computer every time I plug it in. I replicated this behavior with a Kanex eSATA external host, as well as the Thunderbolt 3 dock from Caldigit. In both cases, the computer screen will go blank and the machine will restart. No kernel panic, nothing. Just blank screen and reboot.

I'd be delighted if the controller also offered USB 3 as well as eSATA external interfaces. The cherry on top would be a controller that uses one of those SCSI-1 brackets to allow easy mounting to a multi-bay DVD enclosure (but that's not necessary). However, this has to be a standalone controller, i.e. it should not require being plugged into a PCIe bus to operate.
I misunderstood your original post. ... If you are seeing JBOD and RAID, then I am not exactly sure what your end game is. Speed? Can you run Aja or Blackmagic speed tests and report back your results in your desired RAID configuration? My opinion is port multplier might barely give you a tiny speed boost in RAID. (Maybe in JBOD when accessing all 4 disks simultaneously, but I bet the Finder would hate that.). Let us know your results though.
 


Ric Ford

MacInTouch
OWC just cut the price on their 12-port Thunderbolt 3 dock from $299 to $250, though reviews on Amazon are less than stellar (and Amazon hasn't posted the lower price yet).
 


Ric Ford

MacInTouch
And, for comparison, here are the results for the 2017 MacBook Air's internal SSD (128GB):
Write: 684 MB/s
Read: 1123 MB/s
Here's what I got over Thunderbolt 2 in Target Disk Mode, in my best run:

Write: 230 MB/s
Read: 208 MB/s


That's abysmal and non-sensical - hard drive speed from a 20Gbps interface that should easily support well over 1000 MB/s.

What in the world is going on here? This is not an isolated instance, nor a particular Mac, nor a cable problem, nor contention on another port. It's some fundamental flaw in Apple's Thunderbolt Target Disk Mode. But I don't know what that flaw is or why it has persisted for years!

Trying to rule out any other problem factors, I tried all of these changes without any improvement:
  • very short Thunderbolt 2 cable
  • removing display cable from second Thunderbolt port
  • rebooting
  • connecting the power adapter
Who cares, other than me? Anyone who needs to move a lot of data quickly between Macs.

It's actually faster to read/write the data with a USB drive!
 


Ric Ford

MacInTouch
And, for comparison, here are the results for the 2017 MacBook Air's internal SSD (128GB):
Write: 684 MB/s
Read: 1123 MB/s
Here's what I got over Thunderbolt 2 in Target Disk Mode, in my best run:
Write: 230 MB/s
Read: 208 MB/s
I decided to flip the script, so the 2015 MacBook Pro 15" was in Target Disk Mode with the 2017 MacBook Air connected via a short Thunderbolt 2 cable and running Blackmagic Disk Speed Test (with no other device connected):
Write: 635 MB/s
Read: 400 MB/s
Here's what I get for that same internal drive without Thunderbolt Target Disk Mode crippling it:
Write: 957 MB/s
Read: 1869 MB/s
Note also the bizarre Thunderbolt Target Disk Mode results where writes are faster than reads, the opposite of normal behavior - reads are even more drastically crippled than writes, and I can't think of any logical explanation for that.
 


Here's what I got over Thunderbolt 2 in Target Disk Mode, in my best run:

Write: 230 MB/s
Read: 208 MB/s


That's abysmal and non-sensical - hard drive speed from a 20Gbps interface that should easily support well over 1000 MB/s.

What in the world is going on here? This is not an isolated instance, nor a particular Mac, nor a cable problem, nor contention on another port. It's some fundamental flaw in Apple's Thunderbolt Target Disk Mode. But I don't know what that flaw is or why it has persisted for years!

Trying to rule out any other problem factors, I tried all of these changes without any improvement:
  • very short Thunderbolt 2 cable
  • removing display cable from second Thunderbolt port
  • rebooting
  • connecting the power adapter
Who cares, other than me? Anyone who needs to move a lot of data quickly between Macs.

It's actually faster to read/write the data with a USB drive!
Not that this at all addresses the slow Target Disk Mode conundrum, but have you tried using IP over Thunderbolt to transfer large files? IP wouldn't be the most efficient protocol for such a task, but maybe faster than the Target Disk Mode speeds you are seeing. I've done it in the past, but didn't really pay that much attention to the speed. Pro: doesn't require that you shutdown and reboot your computers. Con: having to see up file sharing, etc. (which you may have already set up for sharing smaller files over your wi-fi or ethernet networks).

 


Ric Ford

MacInTouch
Not that this at all addresses the slow Target Disk Mode conundrum, but have you tried using IP over TB to transfer large files? IP wouldn't be the most efficient protocol for such a task, but maybe faster than the Target Disk Mode speeds you are seeing. I've done it in the past, but didn't really pay that much attention to the speed. Pro: doesn't require that you shutdown and reboot your computers. Con: having to see up file sharing, etc. (which you may have already set up for sharing smaller files over your wi-fi or ethernet networks).
Thanks, I've been wondering about exactly that, but hadn't done it before and didn't know exactly how to go about setting it up, so I hadn't gotten to doing the tests yet. Definitely a good question!

Of course, the whole reason for exploring these options (and the problems I've encountered) is because they theoretically make sense for transferring hundreds of gigabytes from one Mac's internal drive to another Mac's internal drive, especially when migrating from one system to another. Thunderbolt Target Disk Mode should be an ideal solution for this, if it weren't crippled for some unknown reason. I haven't done a network-based Mac migration, so that's another place I have to get up to speed, if Thunderbolt network transfers are faster than the dysfunctional Thunderbolt Target Disk Mode.

I've got a big Samsung T5 as an alternative migration path, if Thunderbolt gets too frustrating with all its problems, and I just booted an old 2011 MacBook Pro off the T5 via USB 2... which was less painful than I expected - not even all that slow!
 


Ric Ford

MacInTouch
I decided to flip the script, so the 2015 MacBook Pro 15" was in Target Disk Mode with the 2017 MacBook Air connected via a short Thunderbolt 2 cable and running Blackmagic Disk Speed Test (with no other device connected):
Write: 635 MB/s
Read: 400 MB/s
Here's what I get for that same internal drive without Thunderbolt Target Disk Mode crippling it:
Write: 957 MB/s
Read: 1869 MB/s
Note also the bizarre Thunderbolt Target Disk Mode results where writes are faster than reads, the opposite of normal behavior - reads are even more drastically crippled than writes, and I can't think of any logical explanation for that.
So, what happens if you connect the blindingly fast internal SSD of a 2018 Mac Mini in Target Disk Mode across the Thunderbolt 3-Thunderbolt 2 adapter and test it from a 2017 MacBook Air?
Write: 440 MB/s
Read: 348 MB/s

Wow. That's pretty special... for a drive that benchmarks like this:
Blackmagic Disk Speed test: 2624 / 2761 (write / read).
Really, Apple? Why isn’t this crippling limitation at least documented if you can’t manage to fix it after years of touting Thunderbolt’s theoretical speed? Performing many times slower than advertised with no explanation isn't too cool.
 


Ric Ford

MacInTouch
So, as expected, the new Mac Mini is quite powerful in CPU operations (single and multi-threaded), while graphics are adequate but unimpressive. What really is impressive is the internal (500GB) flash drive.

AJA System Test Lite says:
Write: 1967 MB/s
Read: 2657 MB/s
Blackmagic Disk Speed Test says:
Write: 1864 MB/s
Read: 2521 MB/s
So, what happens if you connect the blindingly fast internal SSD of a 2018 Mac Mini in Target Disk Mode across the Thunderbolt 3-Thunderbolt 2 adapter and test it from a 2017 MacBook Air?
Write: 440 MB/s
Read: 348 MB/s
Note also the bizarre Thunderbolt Target Disk Mode results where writes are faster than reads, the opposite of normal behavior - reads are even more drastically crippled than writes, and I can't think of any logical explanation for that.
Not that this at all addresses the slow Target Disk Mode conundrum, but have you tried using IP over Thunderbolt to transfer large files? IP wouldn't be the most efficient protocol for such a task, but maybe faster than the Target Disk Mode speeds you are seeing. I've done it in the past, but didn't really pay that much attention to the speed. Pro: doesn't require that you shutdown and reboot your computers. Con: having to see up file sharing, etc. (which you may have already set up for sharing smaller files over your wi-fi or ethernet networks)
Thanks again, Joe, for posting that link. I hooked up the new Mac Mini and the 2015 MacBook Pro with a Thunderbolt cable and Apple's Thunderbolt 3-Thunderbolt 2 adapter, then I enabled file-sharing on the Mac Mini, and, lastly, mounted an HFS+ partition shared via AFP from the Mac Mini's internal drive to run benchmarks from the MacBook Pro.

AJA System Test Lite says:
Write: 1212 MB/s
Read: 355 MB/s
Blackmagic Disk Speed Test says:
Write: 1010 MB/s
Read: 293 MB/s

Again, the Thunderbolt writes are bizarrely faster than reads, which makes no sense. But at least writes are faster over the Thunderbolt network than over Thunderbolt Target Disk Mode, while reads are about the same speed.

The benchmark tests were also odd in that read speeds started out very slowly and then picked up to much faster rates during a single pass.

Let's try it over WiFi for comparison:
Blackmagic Disk Speed Test says:
Write: 2.6 MB/s
Read: 2.1 MB/s

That's really painful...
 


Ric Ford

MacInTouch
Thanks again, Joe, for posting that link. I hooked up the new Mac Mini and the 2015 MacBook Pro with a Thunderbolt cable and Apple's Thunderbolt 3-Thunderbolt 2 adapter, then I enabled file-sharing on the Mac Mini, and, lastly, mounted an HFS+ partition shared via AFP from the Mac Mini's internal drive to run benchmarks from the MacBook Pro.

AJA System Test Lite says:
Write: 1212 MB/s
Read: 355 MB/s
Blackmagic Disk Speed Test says:
Write: 1010 MB/s
Read: 293 MB/s

Again, the Thunderbolt writes are bizarrely faster than reads, which makes no sense. But at least writes are faster over the Thunderbolt network than over Thunderbolt Target Disk Mode, while reads are about the same speed.

The benchmark tests were also odd in that read speeds started out very slowly and then picked up to much faster rates during a single pass.

Let's try it over WiFi for comparison:
Blackmagic Disk Speed Test says:
Write: 2.6 MB/s
Read: 2.1 MB/s
Let's try it over Ethernet, using a Thunderbolt-Gigabit Ethernet adapter on the 2015 MacBook Pro and running an old Ethernet patch cable from that to the Mac Mini's built-in 10Gbps port.

Interestingly, I can mount the Mac Mini's internal APFS volume from the MacBook Pro, which is only running macOS Sierra, because I'm going over the common SMB network file-sharing protocol, so the local file system is irrelevant, and it doesn't matter that macOS Sierra can't mount APFS volumes directly.

AJA System Test Lite says:
Write: 81 MB/s
Read: 87 MB/s
Blackmagic Disk Speed Test says:
Write: 75.6 MB/s
Read: 71.6 MB/s

So, the conclusions are pretty clear here:
  1. High-speed Thunderbolt drives should offer the fastest data transfer in and out of a compatible Thunderbolt-equipped Mac.
  2. Thunderbolt networks are suprisingly faster than Apple's crippled Thunderbolt Target Disk Mode for writing data but similar for reading data.
  3. USB 3 SSDs are likely faster than Thunderbolt Target Disk Mode for transferring data.
  4. Thunderbolt Target Disk Mode is faster for transferring data than Gigabit Ethernet.
  5. USB flash drives are roughly similar (but not identical) to Gigabit Ethernet for file-transfer speed, depending on the particular flash drive.
  6. WiFi data transfer is excruciatingly slow - worse than other local transfer options (but may be fine for a few small files).
 


Let's try it over Ethernet, using a Thunderbolt-Gigabit Ethernet adapter on the 2015 MacBook Pro and running an old Ethernet patch cable from that to the Mac Mini's built-in 10Gbps port.

Interestingly, I can mount the Mac Mini's internal APFS volume from the MacBook Pro, which is only running macOS Sierra, because I'm going over the common SMB network file-sharing protocol, so the local file system is irrelevant, and it doesn't matter that macOS Sierra can't mount APFS volumes directly.

AJA System Test Lite says:
Write: 81 MB/s
Read: 87 MB/s
Blackmagic Disk Speed Test says:
Write: 75.6 MB/s
Read: 71.6 MB/s

So, the conclusions are pretty clear here:
  1. high-speed Thunderbolt drives should offer the fastest data transfer in and out of a compatible Thunderbolt-equipped Mac.
  2. Thunderbolt networks are suprisingly faster than Apple's crippled Thunderbolt Target Disk Mode for writing data but similar for reading data.
  3. USB 3 SSDs are likely faster than Thunderbolt Target Disk Mode for transferring data.
  4. Thunderbolt Target Disk Mode is faster for transferring data than Gigabit Ethernet.
  5. USB flash drives are roughly similar (but not identical) to Gigabit Ethernet for file-transfer speed, depending on the particular flash drive.
  6. WiFi data transfer is excruciatingly slow - worse than other local transfer options (but may be fine for a few small files).
Is AirDrop any different than #6, WiFi data transfer?
 



I am seeking a 5-port eSATA to SATA port multiplier for use inside an external enclosure....
A follow-up, and a couple of thoughts...

The port multiplier from Addonics is working great, though not with the eSATA port inside the Caldigit Thunderbolt 3 dock or the Kanex Thunderbolt 2 adapter. However, DATOptic does offer a $33 USB3-eSATA 'converter' that just works. So that problem is solved, all the DVD drives whrirr away at the same time. Happiness.

Also thought you might be amused to hear that hardware from '90s still works just fine with a 2016 MacBook Pro - however, a long chain of connections was required to connect ancient technology with Apple's more recent wares, i.e:

DGR Technologies
128MB rewritable optical system
|
25-to-50 pin SCSI cable
|
Orange FireWire 400-SCSI adapter
|
FireWire 400-800 cable
|
Thunderbolt 2-FireWire 800 adapter
|
Thunderbolt 3-2 adapter
|
2016 MacBook Pro
I was happy to get 10MB/minute. I still had the cables! It just worked!

Trouble is, you can get the files but unless you have Basilisk, etc. installed, you won't be able to read many of them. Seeing that I haven't bothered in 30 years, it clearly has not been a major priority...
 


Oh, and speaking of target disk mode, I just repurposed a Mac Mini Core 2 Duo (4,1) to run Windows 10 via target disk mode.

I had to follow instructions on the web about how to accomplish this without the help of Boot Camp, since Boot Camp as of OS X 10.8 didn't allow for a Windows 10 installation, only Windows 7.

Then, I discovered that even if I nuked the hard drive in target disk mode and formatted it as a single MS-FAT partition, the Windows installer didn't want to install on it, even after successfully formatting it as a NTFS volume.

Thanks to the great instructions here, along with the GPT fdisk utility by Rod Smith, I was able to fix the issue and install Windows 10 on the machine without errors. It runs just fine. It'll become my principal Arduino IDE machine.

FWIW, I used a physical stick from Microsoft to install Windows 10 Home. I also downloaded the necessary drivers with the help of Boot Camp at the beginning of the process. However, it's entirely unclear to me if Windows ever went looking for them. More likely than not, they may be just for dual-boot machines?

Anyhow, just goes to show how Apple seems to have lost their way. If Microsoft can operate its most recent OS on machines that are ten years old, so should Apple. After all, the hardware entropy that Apple has to deal with is multiple orders of complexity smaller than Microsoft's.
 


The port multiplier from Addonics is working great, though not with the eSATA port inside the Caldigit Thunderbolt 3 dock or the Kanex Thunderbolt 2 adapter. However, DATOptic does offer a $33 USB3-eSATA 'converter' that just works. So that problem is solved, all the DVD drives whrirr away at the same time.
Since the USB-A ports are USB 3.0 and the USB-C ports are USB 3.1, I'd think this would work better:
(It's basically the same price, but USB-C and USB 3.1.)

The main reason I'm looking at that one, personally, is that I'd like to use it with my existing “toaster”-style hard drive dock for rotating backups. I still need to figure out how to house my two spinning disks in a reasonably inexpensive fashion. A few hundred dollars for a Thunderbolt dock seems wrong after complaining about Apple’s RAM-and-SSD prices!

(And in case anyone's wondering, I'm trying to balance the thought of an iMac 21” + external 27” monitor against the Mini with a single 27” or 32” monitor.)
 


Thanks again, Joe, for posting that link. I hooked up the new Mac Mini and the 2015 MacBook Pro with a Thunderbolt cable and Apple's Thunderbolt 3-Thunderbolt 2 adapter, then I enabled file-sharing on the Mac Mini, and, lastly, mounted an HFS+ partition shared via AFP from the Mac Mini's internal drive to run benchmarks from the MacBook Pro.

AJA System Test Lite says:
Write: 1212 MB/s
Read: 355 MB/s
Blackmagic Disk Speed Test says:
Write: 1010 MB/s
Read: 293 MB/s

Again, the Thunderbolt writes are bizarrely faster than reads, which makes no sense. But at least writes are faster over the Thunderbolt network than over Thunderbolt Target Disk Mode, while reads are about the same speed.

The benchmark tests were also odd in that read speeds started out very slowly and then picked up to much faster rates during a single pass.

Let's try it over WiFi for comparison:
Blackmagic Disk Speed Test says:
Write: 2.6 MB/s
Read: 2.1 MB/s

That's really painful...
Lloyd Chambers had done some testing of Thunderbolt networking and was less than impressed. He didn't test it in Target Mode.
macOS Thunderbolt 10 Gigabit Networking Still Not Ready for Prime Time​
 


Amazon disclaimer:
As an Amazon Associate I earn from qualifying purchases.

Latest posts