Samsung's Portable SSD T1 Tested

MojoKid writes The bulk of today's high-capacity external storage devices still rely on mechanical hard disk drives with spinning media and other delicate parts. Solid state drives are much faster and less susceptible to damage from vibration, of course. That being the case, Samsung saw an opportunity to capitalize on a market segment that hasn't seen enough development it seems--external SSDs. There are already external storage devices that use full-sized SSDs, but Samsung's new Portable SSD T1 is more akin to a thumb drive, only a little wider and typically much faster. Utilizing Samsung's 3D Vertical NAND (V-NAND) technology and a SuperSpeed USB 3.0 interface, the Portable SSD T1 redlines at up to 450MB/s when reading or writing data sequentially, claims Samsung. For random read and write activities, Samsung rates the drive at up to 8,000 IOPS and 21,000 IOPS, respectively. Pricing is more in-line with high-performance standalone SSDs, with this 1TB model reviewed here arriving at about $579. In testing, the drive did live up to its performance and bandwidth claims as well.

Comments

no

[TheGratefulNet]

regular ssd, usb3 interface, UASP (scsi over usb, new standard) and you have all the speed of native sata (that the drives can put out) and are still vendor neutral.

I try to avoid samsung products these days. after the fiasco with the evo drives, I'll look for another vendor.

and then there is always the worry that samsung will insert commercials between disk block seeks (inside joke, sorry if that does not make immediate sense to you).

Re:no

[ihtoit]

I'm more concerned with Samsung uploading an incremental mirror of your hard drive to its cloud.

(ditto inside joke thing).

Re:no

[Bengie]

Intel has a bug that makes you lose all of your data, oops. Samsung has a bug that reduces the speed of your drive, then they offer a fix, OMG BURN THEM!

Re:no

[sexconker]

SCSI over USB only really adds queuing, improving speed when many small reads/writes are performed, and you'd need an SSD supporting SCSI and an enclosure/adapter supporting SCSI over USB. Further, for large transfers plain USB 3 is just as fast, while having the benefit of being cheaper, and more readily available and compatible than SCSI over USB. Of course, straight SATA III (via eSATA if you want) is still faster.

USB 3 gets you 5 Gbps and has to be handled by the CPU.
SATA III gets you 6 Gbps without going through the CPU.

USB 3.1 promises to get you 10 Gbps (and lower overhead), but still has to go through the CPU.
And Thunderbolt is just a convoluted and expensive way of piping a limited number of PCIe lanes to a random physical port and requiring the user to buy an expensive cable. 10 Gbps or 20 Gbps. 40 Gbps in the next revision.

SATA Express / M.2 can get you 32 Gbps using 4 PCIe 3.0 lanes or 2 PCIe 4.0 lanes wrapped up in NVMe.
And you can always just throw more PCIe lanes at some controller (on-board or an via a PCIe slot) or some device directly if you want more bandwidth.

USB 3 will be the standard for external shit for a long time. The C connector and USB 3.1 are going to have a hell of a time gaining traction.
For people who want performance, SATA Express / M.2 using NVMe or other direct PCIe solutions win.

Re:no

[cb88]

SCSI over USB isn't exactly all that new its been in Linux since 2012, windows since version Win8 and Mac OS since 10.8. I definitly heard about it back in 2012 maybe even in 2011.

Re:no

[MrBingoBoingo]

A number of 840 evo drives experienced substantial performance degradation much faster than was reasonable. That said I ended up getting an 850 evo recently because with Samsung's output volume I imagine their whiner/[just works] ratio is probably rather favourable to the consumer. In the past I mostly used Intel SSD's to great life span (storage needs began exceeding requirements well before hardware failure). When buying as SSD the big concern is going to be how many is the maker putting out versus how many complaints there are. Of course with the recent resurrection of storage firmware diddling in the news it would make sense to take measures to keep NSAware away from your drives as it can't be good for longevity or performance.

Re:no

[Bengie]

The 840 EVO performance issue was fixed with a firmware update.

Re:no

[thegarbz]

The 840 EVO performance issue was fixed with a firmware update.

Based on my research at www.slashdot.org all that update did was upload your data to Samsung and kill one kitten per 10000 IOPS. I wouldn't trust Samsung to get anything right. If you want some true reliability from an amazing vendor who never does consumers any wrongs, buy OCZ instead.

Re:no

[ihtoit]

even further back. I have three Adaptec SCSI-to-USB adapters - actually physical pin-compatibility adapters. I've had those since probably 2005 or even before. They'll mount on pretty much anything I plug them into, from Windows ME through 7, OSX from Tiger/PPC (the one I've tried it on), and several flavours of Linux from around Knoppix 5.1.1 and I can still read every hard drive I still own from a 10MB 40-pin Winchester through the pile of 500GB Deskstars, several Seagate 9.1GB UW ans a good few 50-pin random and various capacity drives - not forgetting of course, the takep drives, slot loading and cassette DVD/R/RW/RAM drives and my pride and joy of MO gear that still works: a custom cased LS120/Zip100 triple threat (it reads 3.5" floppies, too!). All USB mass storage is really just SCSI layer on the USB stack.

NSA Backdoor preinstalled?

[Anonymous Coward]

NSA Backdoor preinstalled?

Re:NSA Backdoor preinstalled?

[Anonymous Coward]

https://www.reddit.com/r/news/comments/2w4ihb/kaspersky_labs_has_uncovered_a_malware_publisher/

Danger of SSDs

[bradgoodman]

for a ton of technical reasons I won't get into right now (remapping/wear leveling) SSDs aren't usually able to handle power faults like regular HDs. Too often, taking an unexpected power hit can easily result of massive amounts of lost data, or even loss of the device itself. I've seen this happen at least 20 times. Thete are allegedly some "enterprise grade SSDs" which may or may not mitigate this issue. I'm tired of seeing articles citing all kinds of performance tests that go into absolutely no detail on if you are going to lose all your data the next time you lose power, or have to force-off your laptop because it locked-up on you.

Re: Danger of SSDs

[dpidcoe]

I thought this was the reason a lot of SSDs now have a collection of capacitors to finish out the writes with in the event of a power loss?

Re:Danger of SSDs

[Anonymous Coward]

That is luck...

I worked on a project about 15 years ago that shipped about 8meg of flash on an onboard truck device.

We had about 20% fail rate per year (yeah it was very expensive). Pretty much every one of them was returned because of non bootup, they were corrupt. Reflash it and it would come back to life. We would ship it back out as a refurb and it would be back in 2-3 months. The only way we could reproduce it in the lab was power flickage. Not completely off (though we suspected that would happen no one could manage to pull it off). When a truck starts up it bounces between 1/2v to 16v. Bellow 4v this particular chip would not write or would write garbage. It would even get worse as the more you wrote to it. As flash gets slower the more you write to it. Turned out the super cap we had put in to mitigate the issue was not enough to account for age... So we went from 1-2 returns a month to hundreds. We had to open every one up and replace the chips with newer ones for it to be 'fixed' but that would only last for about 3 years.

For it to be corrupt you have to be writing something at the same time. If you corrupt a cell in the middle of your page file or a temp file in the middle of a power off would you notice? Because when you come back those files by definition are junk and free game to throw away.

Many SSDs out there will finish the current page it happens to be writing. That is about all they guarantee. Many do not even do that and just go off.

Do I use an SSD? Hell yeah. The speed increase is worth it. 500 meg a second is well worth it. I just make sure I have a full drive image.

Re: Danger of SSDs

[Bengie]

I would like to more about this topic. According to Samsung, the 850 drives reserve a portion of the drive to use as SLC allowing the DRAM to be quickly written to the SLC and allows the drive to slowly write out to the MLC. From the sounds of it, Samsung is less concerned about the drive internal state getting corrupted and committed data being lost and more concerned with in-flight data that hasn't been written being committed.

I'm also not concerned with losing data during a write, I just don't want to lose my drive. I wish there was more information on this topic.

Re:Danger of SSDs

[mlts]

I wonder if the latest generation of filesystems like ZFS, btrfs, and ReFS would be useful, so a corrupt file that wasn't completely written would be detected by the FS during a background scrub or garbage collection task. With RAID-Z, the corruption can be found. Z2, the corruption likely can be fixed.

Re: Danger of SSDs

[Bengie]

SSDs with "power loss protection" store enough power to write out all of their cache, which is something like 1GB now days. Like we've mentioned, we don't care about caches not being flushed, but how to the internal mapping tables hold up without "power loss protected". My hope would be that modern controllers can handle keeping internal state and just screw the data in cache.

I was reading about Samsung's "RAPID Mode" that uses system memory as a write cache to speed up writes to the SSD. One of the topics about "RAPID Mode", which is even more sensitive to power loss because of increase caching, is that it handles power loss "well". They have done extensive testing with "RAPID Mode" and power loss. I figure if they can offer 10 year warranties and feel confident about these issues, I'll trust them until proven otherwise. They have a great track record. I still wouldn't put all of my eggs in one basket.

Re:Danger of SSDs

[Curtman]

Intel 5400 RPM enterprise SSDs are the industry leader. I've never had issues with the ones I have.

I keep mine on a 33 1/3 RPM turntable. Less centrifugal force. Damn cables keep getting tangled though.

WTF?

Re:Danger of SSDs

[jtownatpunk.net]

I overclock mine to 78 RPM.

Re:NAND is for chumps

[TechyImmigrant]

A TB has already been done in SLC, that isn't hard to do... it is however enterprise only and very expensive, thus not of much interest to your average user...

It isn't going to be something that you see for your personal use.

Why not? I want multi TB SSD in full size HDD boxes, that can go into standard PC drive slots. Drives fail, even SSDs. I want to mirror them. 1TB MLC SSDs exist. Put them in a bigger box.

Re:NAND is for chumps

[Bengie]

Modern MLC drives are able to handle nearly 1PB of writes. I'm not sure what more you want from a 120GB SSD.

Re:NAND is for chumps

[Bengie]

You can purchase two 1TB SSDs and mount them in this http://www.newegg.com/Product/...

Re:NAND is for chumps

[mlts]

The price is dropping. I'm seeing MacBook Pros ship with 1TB of SSD. It only is a matter of time before external SSDs become the storage medium of choice, just like USB flash drives are for small scale storage.

As for HDDs, I can see them winding up being re-engineered to be more for archival and backup storage as opposed to the role an external HDD does now.

Re:What Is...

[fuzzyfuzzyfungus]

Given the cost per gram of contemporary flash memory, I'm going to guess "Really, really, heroically expensive".

Re:NAND is for chumps

[MrBingoBoingo]

This. SLC offers a "promise" of greater longevity with lots of write cycling. Still when most SSD's fail they die at the controller.

Re:NAND is for chumps

[WuphonsReach]

I suspect the market isn't there yet for 4 TB SSD drives... and it wouldn't require a 3.5" drive case, you could fit 4 TB of NAND easily in a 2.5" drive case (or even less).

Well right now, the 1TB enterprise quality SSDs have dropped below $900 (5 year warranty, super-caps, etc.). They're quickly edging out the 15k RPM SAS drives.

Consider that if I need X IOPS and a TB of capacity, I can either put 2x1TB SSD into a server and spend about $1800-ish. Or I can buy a more expensive RAID controller and try to put together half a dozen to a full dozen 15k RPM SAS drives. Using SSD means less drives, less power, less heat, smaller server footprint - same or better IOPS (usually 10x better).

The magical price point for enterprise storage drives is about $300-$600 per drive. Business IT won't blink at spending $300-$600 on a single drive, especially if it reduces their spindle count and increases performance. Something in the $1500-$2000 range tends to be a rarer purchase.

Re:People are looking at the wrong specs

[hankwang]

Do you have the same complaint about car top speeds in mph or km/h rather than min/km or min/mi? And your car speedometer specifies infinity while idling at the traffic light?