This is about $200-300 more expensive than a 2 TB SATA SSD. I guess the price is fair if you absolutely need that capacity and speeds higher than what USB 3.0 gives.
Also, at $800 this SSD costs more than top-of-the-line GPUs from AMD and NVidia.
What has this got to do with GPUs? Why does it seem as though everyone on this site is only interested in graphics cards? The product being reviewed here is not even remotely aimed at the gaming market. Have I missed something, or am I finally the only normal person on a forum?
Both SanDisk and Samsung have smaller capacity, lower priced versions of their drives that would be of interest. For example on Amazon SanDisk has a 960GB version for $500 which sounds like just 1 960 GB instead of two in RAID 0 (which may also solve the issue of no trim if its a single drive rather than RAID 0???). Samsung has: 250 GB T3 for $129, 500 for $197 and 1 TB for $427. The 500 GB actually seems pretty reasonable to me as its big enough to fit several VM's of the size I'd use it for and the price is still relatively low.
The Mushkin Enhanced Reactor TC 2.5" 2 TB SATA III MLC Internal Solid State Drive (SSD) MKNSSDRE2TB-TC from Amazon costs $500. Place this in a thin $20 USB 3.0 enclosure. You'll save $300 over the SanDisk and Samsung external 2TB SSDs.
Sounds like a decent way of increasing the iops/storage when running a bunch of virtual machines on a laptop or even a desktop machine and taking them with you whereever you go.
Not free for sure, but compared to the price of internal storage options and impossibility of even having 2TB in most laptops, this is not bad.
External storage for a virtual machines seems like a good use case for the Extreme 900. I can't think of too many others that would justify the fairly steep MSRP vs storage capacity that can readily benefit from the added performance.
Do those $10 enclosures support RAID 0? I'm sure there are some on the market, but to support RAID 0, USB 3.1 gen2? Doubt there are many if any on the market.
I've got this 2 bay 3.1 gen2 from Startech. Have two 480gb Sandisk Ultra II SSD's running RAID 0. I get 420 MB/s read and write with my Macbook Retina. I'm on the road, so I haven't tried them on my desktop, with true 3.1 gen2 support yet.
I'd rather have a boring enclosure design that is as small and utilitarian as possible. No need for extra fins and doodads. Just do its job and stay out of the way.
I wonder if we'll see at some point external enclosures capable of supporting PCIe SSDs that can saturate 10Gb/s and/or even use Thunderbolt 3 to transfer data up to 40Gb/s
"with a peak power consumption of around 8.1W. SanDisk suggests using USB ports capable of delivering up to 10W of power for optimal performance. It is obvious that using a port capable of delivering only the usual 5W will heavily hamper the performance of the unit. "
I'm sorry but this is an unacceptable ending to a pretty good review. There is no such thing as a "usual 5W USB port". A USB2 port will give you 2.5W, a USB3 port will give you 4.5W. Hoping that you'll be lucky and get more is a recipe for tears.
Next, since SANDisk provide an A-connector, what actually HAPPENS when you use a 2.5 or 4.5W port? I'm somewhat dubious of the claim that it will just "heavily hamper the performance". At the very least I'd like to see some testing of this, with verification that the device does not simply randomly disconnect when it wants to power-draw beyond the allowed limit. Even assuming it is well-enough engineered to run properly on these lower power-draws, how much of a performance hit are we talking?
This is part of a larger problem that, especially in the context of external drives (and, to some extent, WiFi equipment) AnandTech lives in blissful ignorance of the real world. Sure your labs don't contain a single devices manufactured before January 2016, but in the real world, a substantial use case for external USB drives involves their being swapped between different machines. THIS is why it REALLY matters to know how well they can handle being connected to older ports.
This is not just some weird corner case, like a person complaining "well you didn't test that new SATA SSD drive when you connect it to 2007 SATA-66 box, and I really care about that situation"; it gets to the actual fitness of purpose for I would guess many, if not a majority, of the use cases of this sort of external USB drive.
All hyperbole from name99 aside; some degree of benchmarking at lower power levels needs to be done for all high power usb flash devices for the next few years. Not necessarily the full suite; but one or two tests suitable to characterize what the fallback performance levels will be for people not having the latest and greatest hardware to plug them into.
Which is why the SIIG USB 3.1 Gen 2 single bay enclosure comes with a dual host port cable, Type A on both ends.
I have a 480 GB Patriot Blast SSD in it and it does a pretty nice job of saturating the SATA bus. Things should get more interesting if and when somebody produces an SSD controller that avoids the SATA bottleneck and taps into USB 3.1 fully.
The SATA-IV standard should support a minimum clock rate of 8 GHz (to synchronize it with the 8G clock already supported by the PCIe 3.0 standard), and it should also support the 128b/130b "jumbo frame" also in the PCIe 3.0 standard. PCIe 4.0 will oscillate at 16 GHz, and SATA-V should also sync with that faster clock. The fact that USB 3.1 increased the clock rate to 10G and adopted a 128b/132b jumbo frame is proof that the SATA standards group have morphed into snails marooned on dry ice.
We’ve updated our terms. By continuing to use the site and/or by logging into your account, you agree to the Site’s updated Terms of Use and Privacy Policy.
Facebook
Twitter
RSS
31 Comments
Back to Article
Chaitanya - Monday, April 11, 2016 - link
Would have liked to see internals of the drive.Ethos Evoss - Tuesday, April 12, 2016 - link
What bout TRIM ? SSDs via usb doesn't support it .. so SSD will degrade ...jji7skyline - Thursday, April 14, 2016 - link
Don't some SSD controllers have TRIM function built in?Ethos Evoss - Thursday, April 14, 2016 - link
Not reallyjameskatt - Saturday, May 7, 2016 - link
Yes. The SSDs from OWC don't need TRIM.chlamchowder - Monday, April 11, 2016 - link
This is about $200-300 more expensive than a 2 TB SATA SSD. I guess the price is fair if you absolutely need that capacity and speeds higher than what USB 3.0 gives.Also, at $800 this SSD costs more than top-of-the-line GPUs from AMD and NVidia.
MattMe - Monday, April 11, 2016 - link
What has this got to do with GPUs?Why does it seem as though everyone on this site is only interested in graphics cards? The product being reviewed here is not even remotely aimed at the gaming market.
Have I missed something, or am I finally the only normal person on a forum?
nandnandnand - Monday, April 11, 2016 - link
It's overpriced storage for a niche market.dontlistentome - Tuesday, April 12, 2016 - link
Probably not overpriced for the niche it's aimed at....Mattly - Monday, April 11, 2016 - link
LOL if your the only person on the forums that you consider normal perhaps you should question why you are the only one?MattMe - Tuesday, April 12, 2016 - link
The comment about being normal was written tongue firmly in cheek.pav1 - Tuesday, April 12, 2016 - link
Normal people don't post on forums.Ratman6161 - Monday, April 11, 2016 - link
Both SanDisk and Samsung have smaller capacity, lower priced versions of their drives that would be of interest. For example on Amazon SanDisk has a 960GB version for $500 which sounds like just 1 960 GB instead of two in RAID 0 (which may also solve the issue of no trim if its a single drive rather than RAID 0???). Samsung has: 250 GB T3 for $129, 500 for $197 and 1 TB for $427. The 500 GB actually seems pretty reasonable to me as its big enough to fit several VM's of the size I'd use it for and the price is still relatively low.ganeshts - Monday, April 11, 2016 - link
The 960GB version of the Extreme 900 is 2x 480GB in RAID-0jameskatt - Saturday, May 7, 2016 - link
The Mushkin Enhanced Reactor TC 2.5" 2 TB SATA III MLC Internal Solid State Drive (SSD) MKNSSDRE2TB-TC from Amazon costs $500. Place this in a thin $20 USB 3.0 enclosure. You'll save $300 over the SanDisk and Samsung external 2TB SSDs.zepi - Monday, April 11, 2016 - link
Sounds like a decent way of increasing the iops/storage when running a bunch of virtual machines on a laptop or even a desktop machine and taking them with you whereever you go.Not free for sure, but compared to the price of internal storage options and impossibility of even having 2TB in most laptops, this is not bad.
BrokenCrayons - Monday, April 11, 2016 - link
External storage for a virtual machines seems like a good use case for the Extreme 900. I can't think of too many others that would justify the fairly steep MSRP vs storage capacity that can readily benefit from the added performance.ImSpartacus - Monday, April 11, 2016 - link
It's sad that we have to say, "usb 3.1 gen2" now...Flunk - Monday, April 11, 2016 - link
USB 3.1 Gen 1 is such a scam.Gunbuster - Monday, April 11, 2016 - link
I would have thought $800 would afford you a enclosure design slightly more exciting than what can be had on amazon or eBay for $10Holliday75 - Monday, April 11, 2016 - link
Do those $10 enclosures support RAID 0? I'm sure there are some on the market, but to support RAID 0, USB 3.1 gen2? Doubt there are many if any on the market.littlebitstrouds - Monday, April 11, 2016 - link
I've got this 2 bay 3.1 gen2 from Startech. Have two 480gb Sandisk Ultra II SSD's running RAID 0. I get 420 MB/s read and write with my Macbook Retina. I'm on the road, so I haven't tried them on my desktop, with true 3.1 gen2 support yet.http://www.amazon.com/StarTech-com-10Gbps-External...
ganeshts - Monday, April 11, 2016 - link
Nice one.. it is not $10, but $95, but a good solution for sure.Unfortunately, it is not bus-powered or as compact as the Extreme 900. That said, it is definitely more flexible with the configurable RAID levels.
jameskatt - Saturday, May 7, 2016 - link
I'd rather have a boring enclosure design that is as small and utilitarian as possible. No need for extra fins and doodads. Just do its job and stay out of the way.digiguy - Monday, April 11, 2016 - link
I wonder if we'll see at some point external enclosures capable of supporting PCIe SSDs that can saturate 10Gb/s and/or even use Thunderbolt 3 to transfer data up to 40Gb/sname99 - Monday, April 11, 2016 - link
"with a peak power consumption of around 8.1W. SanDisk suggests using USB ports capable of delivering up to 10W of power for optimal performance. It is obvious that using a port capable of delivering only the usual 5W will heavily hamper the performance of the unit."
I'm sorry but this is an unacceptable ending to a pretty good review.
There is no such thing as a "usual 5W USB port". A USB2 port will give you 2.5W, a USB3 port will give you 4.5W. Hoping that you'll be lucky and get more is a recipe for tears.
Next, since SANDisk provide an A-connector, what actually HAPPENS when you use a 2.5 or 4.5W port? I'm somewhat dubious of the claim that it will just "heavily hamper the performance". At the very least I'd like to see some testing of this, with verification that the device does not simply randomly disconnect when it wants to power-draw beyond the allowed limit. Even assuming it is well-enough engineered to run properly on these lower power-draws, how much of a performance hit are we talking?
This is part of a larger problem that, especially in the context of external drives (and, to some extent, WiFi equipment) AnandTech lives in blissful ignorance of the real world. Sure your labs don't contain a single devices manufactured before January 2016, but in the real world, a substantial use case for external USB drives involves their being swapped between different machines. THIS is why it REALLY matters to know how well they can handle being connected to older ports.
This is not just some weird corner case, like a person complaining "well you didn't test that new SATA SSD drive when you connect it to 2007 SATA-66 box, and I really care about that situation"; it gets to the actual fitness of purpose for I would guess many, if not a majority, of the use cases of this sort of external USB drive.
ganeshts - Monday, April 11, 2016 - link
No disconnects. Checked with a traditional USB 3.0 port. Some sort of throttling going on, which reduces the speed.DanNeely - Tuesday, April 12, 2016 - link
All hyperbole from name99 aside; some degree of benchmarking at lower power levels needs to be done for all high power usb flash devices for the next few years. Not necessarily the full suite; but one or two tests suitable to characterize what the fallback performance levels will be for people not having the latest and greatest hardware to plug them into.epobirs - Monday, April 11, 2016 - link
Which is why the SIIG USB 3.1 Gen 2 single bay enclosure comes with a dual host port cable, Type A on both ends.I have a 480 GB Patriot Blast SSD in it and it does a pretty nice job of saturating the SATA bus. Things should get more interesting if and when somebody produces an SSD controller that avoids the SATA bottleneck and taps into USB 3.1 fully.
TheUsual - Monday, April 11, 2016 - link
No 'The'? I don't think every device deserves a 'The' at the beginning of the review, just the very important, well known devices.MRFS - Saturday, April 16, 2016 - link
The SATA-IV standard should support a minimum clock rate of 8 GHz(to synchronize it with the 8G clock already supported by the PCIe 3.0 standard),
and it should also support the 128b/130b "jumbo frame" also in the PCIe 3.0 standard.
PCIe 4.0 will oscillate at 16 GHz, and SATA-V should also sync with that faster clock.
The fact that USB 3.1 increased the clock rate to 10G and adopted a 128b/132b jumbo frame
is proof that the SATA standards group have morphed into snails marooned on dry ice.