Microsoft Testing Windows 10 Feature That'll Detect If Your SSD Is Failing (reuters.com) 39
Microsoft is testing a new feature for Windows 10 that will alert you if your SSD drive is failing. Microsoft is also testing an update to Your Phone that will allow it to work with multiple devices. PCWorld reports: Both features arrived as part of Windows 10 Insider Build 20226 for the Dev Channel, Microsoft's laboratory for future features. The Dev Channel is truly experimental, meaning that these two new features may or may not become official features of the operating system. Fortunately, both are straightforward. An aftermarket SSD may ship with utility software that monitors an NVMe SSD drive's health, but Windows itself does not monitor the drive. In this test feature, Windows 10 will add NVMe SSD drives to its monitoring processes, and let you know if it's about to fail. If you then go into the Windows 10 Settings menu for Storage, you'll see that the SSD drive in question is listed as unreliable. In that case you're advised to back up everything. "Attempting to recover data after drive failure is both frustrating and expensive," Microsoft said in a blog post. "This feature is designed to detect hardware abnormalities for NVMe SSDs and notify users with enough time to act. It is strongly recommended that users immediately back up their data after receiving a notification."
SMART information? (Score:4, Insightful)
So will it just use SMART information and notify you?
By the way the link is probably not going to the right article, it points to some DOJ vs Google article
Re: (Score:3)
I suspect this article is going to attract almost 100% of either "that's a GREAT idea!" or "doesn't SMART already do that??"
Re: (Score:2)
What I want to know is that it's actually working -- where's the fake-out EICAR-like option that detects when power-on-hours changes and alerts me? Maybe the daemon isn't running, maybe it's not reading ALL drives, maybe I don't know what the pop-up looks like, is there enough text and warnnigs so that Grandma knows to call her grandson about it or is "just another pop-up to ignore?"
Even better, does it include an 800-number that I can call to
Re: (Score:2)
It will probably be some 6 second evanescent balloon saying "Caution: Contact your System Administrator"
Re: (Score:2)
It will probably be some 6 second evanescent balloon saying "Caution: Contact your System Administrator"
No, it will be a dialog window that steals focus while you are typing so all you know is that there was some kind of message. At least that is what happens when a drive starts reporting bad sectors.
Re: (Score:1, Insightful)
BeauHD showing his level of excellence once again!!!
Re: (Score:3)
Re: (Score:2)
they want to move you to windows only SSD's with (Score:2)
they want to move you to windows only SSD's with build in windows SPEED LOAD TM
Re: SMART information? (Score:1)
No. It will just run a defrag and surface scan in the background and after the 100th run notify you that your ssd is starting to degrade.
Re:SMART information? (Score:4, Interesting)
> So will it just use SMART information and notify you?
Doubtful. SMART isn't reliable, especially because its reliable operation goes against the manufacturers' warranty incentives.
Windows being Windows, it can keep a histogram in memory about read and write performance and notice when the drive starts pleading for its life with screams of latency.
There's an exponential curve when an SSD is going to die. As much as the nerds here are all, "Linux wuz here," I am not aware of a kernel module and userland daemon pair existing yet. Pointers welcome by everybody who runs a NAS or ten.
Plain SMART (Score:5, Insightful)
> So will it just use SMART information and notify you?
Doubtful. SMART isn't reliable,
Smart is reliable, as long as you're not using some cheap no-name Asian crap that you bought over Amazon.
Some datacenter companies like Backblaze (or at some point in time even Google) do publish their stats.
For SSDs, the one that seems to be the most useful are monitoring 177 Wear Leveling Count (wear) and 187 Uncorrectable Error Cnt (failure)
(for HDD that would be 197 and 198).
especially because its reliable operation goes against the manufacturers' warranty incentives.
Nope. Actually goes *for manufacturers' best interest* if smart works reliably:
- IT industry will be happy if smart can alert of incoming failure before some catastrophic incident ("we better replace this drive as it is going to be failing soon" is much better than "oh, shit! we lost terrabytes of data due to dead drives !" *), which will translate into more buys from that peculiar manufacturer.
- Manufacturer has a small financial direct interest: drives getting replaced *before* they failed is going to translate into earlier buys and therefore more frequent buys.
Also, warranty usually concerns, the other end of the curve (see below).
Windows being Windows, it can keep a histogram in memory about read and write performance and notice when the drive starts pleading for its life with screams of latency.
Nope. That's a very poor indicator: too much flase positive.
Latency is going to go to the toilet whenever there's write amplification as part of the normal read-erase-write cycles due to poor filesystem blocks placements on the SSD's erase blocks, or crappy support of TRIM from the OS, or as part of the normal wear-levelling procedure. Or a bad firmware causing more re-reads due to cell decay (see the infamous bug on early Samsung 840 EVO)
Unless you're sure Windows' handling of block alignement and TRIM support is perfect (I haven't touched windows for years, but I seriously doubt), I wouldn't definitely count on that.
There's an exponential curve when an SSD is going to die.
Nope. It's bathtube-shaped. Warranty is for handling the early fail. The current subject is about long age failure at the other end of the curve, about a decade later.
As much as the nerds here are all, "Linux wuz here,"
Yes, I was using smartctl a decade and half ago.
I am not aware of a kernel module
...smart passing is part of the standard sata/sas/pata/scsi vanilla kernel driver, it's even supported for several USB mass storage controles (both in MSD and UAS modes)...
and userland daemon pair existing yet.
...smartd in the smartctl package (standard in any distro under the sun), will happily monitor any value you ask it to and write you e-mail if needed.
Pointers welcome by everybody who runs a NAS or ten.
If you NAS software doesn't support smartd, you should throw it away and stop buying cheap noname asian crap.
---
(*) though any SSD worth it salt should fail gracefully into read-only mode (in my experience with Samsung, Intel SSD and a few other flash media (Transcend uSD)), so you could still salvage data, even more so if your filesystem is Copy-on-Write or Log-Structured (Btrfs, Zfs, F2FS, etc.)
Re: (Score:3)
Doubtful. SMART isn't reliable, especially because its reliable operation goes against the manufacturers' warranty incentives.
Seagate offer included data recovery with their included long-term warranty for an enterprise drive, which only costs marginally more than the cheaper variants. Do they want to be making good on that data recovery offer, or just replacing a pre-fail disk for the customer? How about when HP and other companies are involved further down the chain, off the back of those guarantees?
Windows being Windows, Microsoft should be recommending continuous online backups and at a minimum, weekly (if not daily) offline b
Data recovery? Data recovery!? (Score:2)
I already have my data all backed up.
About Windows clean install with all of my apps recovered, bro?
Re: (Score:2)
Looking at TFA it does appear to just be using SMART data. When values get near end of life it warns you. No idea why they are saying it's designed for NVMe SSDs, it should work with most drives and is something Windows is sorely lacking.
They've been pushing betas in their updates... (Score:5, Funny)
Re:They've been pushing betas in their updates... (Score:4, Insightful)
Yeah, "It is strongly recommended that users immediately back up their data after receiving a notification." should be:
"It is strongly recommended that users constantly back up their data."
First link is broken (Score:1)
A Small Word of Advice (Score:2)
Re: (Score:2)
I'm not a fan of hard drives for long term storage, platter or even SSD. The best bang for the buck storage wise is LTO-4 tape drives. Bought a used drive for $40 (FC drives are dirt cheap) and tapes can be had for $10 each or so if you keep an eye out.
Re: (Score:2)
An LTO-4 tape holds 800 GB. They like to claim 1.6 TB, but that's with compression. A 1 TB hard drive is something like $40. For my backups, for years the cheapest and easiest thing to back a hard drive up with is another hard drive. This may not scale up well for massive amounts of data where tape may still have an edge, but works well at home. Plus I don't have to deal with tapes and a USB hard drive can be connected to about anything.
I am a bit disappointed by E-SATA becoming scarce though. USB 3.m
Has Microsoft developed a buy (Score:3)
It's the new MS Cloud push.... (Score:4, Funny)
Microsoft has detected that your SSD is failing and you are about to loose all of your data, for just $100 per GB you can back it up the Microsoft Azure cloud now.
SMART (Score:2)
Linux, SMART, 18 years ago.
https://en.wikipedia.org/wiki/... [wikipedia.org]
https://www.smartmontools.org/ [smartmontools.org]
Will they add tmpfs next? Linux, 19 years ago.
https://en.wikipedia.org/wiki/... [wikipedia.org]
Re: (Score:3)
Yah, exactly.
OTOH, you have to enable smartd on most distro's same as you have to install the stupid samsung/etc tools on windows. The latter will also tell you when your disk is failing, as will some BIOS's.
Re: (Score:2)
>"OTOH, you have to enable smartd on most distro's same as you have to install the stupid samsung/etc tools on windows."
While that is true,
1) On many distros it is installed and on by default. CentOS, for example.
2) SMART is the same for all distros and all hardware.
3) SMART it is not some foreign addon you have to go find and download and/or install from somewhere.
4) And it doesn't include anything else that you might not know about and not want.
5) SMART is the same license class as the rest of the OS
Re: (Score:2)
Re: Dell! (Score:1)
Is This About That Version 2004 SSD Bug? (Score:2)
Speedfan had this capability for years now (Score:2)
Re: (Score:2)
Maybe they should just install three, and use a best-of algorithm on the outputs.
So a decade or two late, as usual (Score:2)
It is really baffling how MS always manages to be really late to the game.
This was implemented 20+ years ago. (Score:3)
The detection is already in place, for years....perhaps it is time to change the background color from blue though?
And if not... (Score:2)
Re: And if not... (Score:1)
You beat me to it... Defrag and scandisk surface scan running in background.
Windows Causes Unnessary SSD Wear (Score:1)
The usual nuclear option (Score:2)
Given that the whole point of an OS is to manage system resources and protect the integrity of your data in a filesystem, every OS should have a robust, incremental backup solution pre-installed and immediately obvious.
Even before "the cloud" became a thing, almost every OS vendor preferred you to get/buy some 3rd-party product to do backups, and manufacturers just told you to do a factory reset if you ran into problems. Maybe the reason people don't do proper backups is NOT due to user error, but because