Slashdot Mirror
Intel Launches SSD 750 Series Consumer NVMe PCI Express SSD At Under $1 Per GiB
MojoKid writes Today, Intel took the wraps off new NVMe PCI Express Solid State Drives, which are the first products with these high speed interfaces, that the company has launched specifically for the enthusiast computing and workstation market. Historically, Intel's PCI Express-based offerings, like the SSD DC P3700 Series, have been targeted for datacenter or enterprise applications, with price tags to match. However, the Intel SSD 750 Series PCI Express SSD, though based on the same custom NVMe controller technology as the company's expensive P3700 drive, will drop in at less than a dollar per GiB, while offering performance almost on par with its enterprise-class sibling. Available in 400GB and 1.2TB capacities, the Intel SSD 750 is able to hit peak read and write bandwidth numbers of 2.4GB/sec and 1.2GB/sec, respectively. In the benchmarks, it takes many of the top PCIe SSD cards to task easily and at $389 for a 400GB model, you won't have to sell an organ to afford one.
"High Endurance Technology (HET) enables the DC P3700 Series to achieve 1.7TB drive writes per day over a 5 year drive life" "Life Expectancy 2 million hours Mean Time Between Failures (MTBF), 230 years"
From the review:
"We should also note the the SSD 750 Series comes with a 5 year limited warranty and its endurance is rated for 70GB of writes per day, with a total of 219TB written and a 1.2 million hour MTBF or meantime between failures. "
And by looking at some of the SSD endurance tests, I'd be surprised if this card can't beat 1-2PB before dying.
Hopefully Intel didn't add a suicide option into the firmware, like they did with the 335 SSD. As soon as the counter hits 0%, don't reboot or it's a brick. Doesn't matter if it has lots of spare sectors available for replacement, it's going to kill itself.
One step behind bleeding edge is the sweet spot for me. The last gaming rig I built is approaching 3 years and it's still going strong. The only bleeding edge part was the X79 Extreme 11 motherboard. I built it with one of the 750 gig Seagate hybrids which was later replaced with one of their 2tb hybrids. Works plenty fast for me. When I'm gaming, the next level generally finishes loading before the cut-scene is done so faster load times wouldn't make any difference.
From what I've seen, the standard reaction of SSD firmware when write integritiy can't be guaranteed is "commit suicide".
That was true of some of the OCZ Vertex series and other make/models. But this last generation of SSDs seemed to have made that a rarity even under the most extreme conditions.
In the case of this new Intel PCIe SSD card, I believe it has enough capacitance to commit a complete transistor write upon system power failure. As for the lost data for non-commited data?? Well, you're are running a journaling file system, yes? At least the volume won't get corrupted. But anyways yeah, seems like a solid drive you can rely on. Time will tell of course.
Life is not for the lazy.
You're also using 5 (2 per U) drive slots vs 1 (10 per U). And assume that your raid controller can push to the drives at pcie speed. Raid controllers aren't that fast, even from expensive manufacturers chips push the boundaries at 6Gbps and ~100,000 IOPS for the entire array.
Custom electronics and digital signage for your business: www.evcircuits.com
Those improvements are not necessary to reach the full speed of this drive, at 440K IOPS. In my own tests I've even seen a FusionIO drive hit 8GB/s under the old RHEL6 2.6.32 kernel. This new drive is at an amazing price/performance spot, but it's not exploring the upper limits of where the Linux kernel is shooting at.
Handling power off issues is a different problem. What the GP was referring to is how drives will fail spectacularly in the face of anything seen as corruption. You can see some examples in some longevity failure tests.
The problem in those cases was wearout, but the way that happens is scary. Let's say there's a bug in the firmware that causes a write to fail for no good reason. It's quite likely that the drive will kick into a mode where it doesn't trust itself anymore. And the way that will play out on most SSDs, the drive will shut itself down at the firmware level, so it isn't even picked up by the BIOS on boot anymore. What people would expect is getting read-only behavior there; instead they will find everything gone. And unlike most catastrophic spinning drive failures, you could easily hit the same bug that wipes out your data on both halves of a RAID-1 pair at the same time.
Fail spectacularly is a vague term IMO. What were talking about is when the Intel firmware has determined that the SSD is in failure it will allow the drive to boot in a read-only state once. After you shut the power off once receiving the warning the drive commits suicide and will no longer boot or respond, in other words it bricks itself at the firmware layer and there is NO recovery.
What I'd argue is the correct failure mode is boot in read-only and warn that power loss will result in data loss but continue to boot in read-only format with a warning at each boot that files may be corrupt or lost. The intentional bricking aspect is just bad design IMO. The data you need to access could be on a part of the drive that's perfectly fine, in addition you may get the data warning at a time and place where it's simply not feasible to backup everything.
I completely disagree with Intel's failure model and think it's beyond stupid. It should warn the user of corruption and data loss but continue to boot. That way if the person is off somewhere they can backup critical files to either the cloud or a thumb drive and try to recover the non-critical data when they get back. Intentional bricking is just stupid.