Toshiba Publishes Full List of Its Drives Using Slower SMR Technology (blocksandfiles.com) 90
"Toshiba has just published a full list of all the consumer HDDs in their lineup that use SMR (shingled magnetic recording) technology," writes Slashdot reader williamyf. "This comes after the whole submarine consumer SMR HDDs fiasco, and fresh on the heels of Western Digital publishing a full list of all their consumer HDDs using SMR. With this, Seagate is the only HDD vendor which has not yet published a full list of their consumer HDDs using SMR." Blocks and Files reports: Toshiba uses SMR technology -- previously undocumented -- in several desktop drives and in some video surveillance HDDs: P300 6TB, P300 4TB, DT02 6TB, DT02 4TB, DT02-V 6TB and DT02-V 4TB. Certain notebook PC, game consoles, and external consumer drives also use SMR: L200 2TB, L200 1TB, MQ04 2TB and MQ04 1TB. The company does not use SMR in the N300, a NAS drive intended for the consumer market -- unlike Western Digital which uses SMR in some low-end WD Red NAS devices.
Not a dupe... (Score:2)
This sounds a lot like a recent story about Western Digital doing the same thing.
Re:Not a dupe... (Score:5, Informative)
This sounds a lot like a recent story about Western Digital doing the same thing.
Original poster here, following this closely since the news broke.
TL;DR: Not a dupe. Just a follow-up.
All three HDD manufacturers (Western Digital, Seagate and Toshiba) where submarining SMR HDDs in consumer HDDs, without saying so in the driver sticker, or the driver brochure, or the driver spec. sheet.
But Western Digital made the mistake of using SMR on some consumer NAS drives, where a bunch of clever people (among them Allan Brown, and Chris Mellor) realized something was amiss. After some sleuthing, by a lot of people, the cat was out of the hat.
In the begining, WD denied any wrongdoing, and doubled down on the line of "you do not need to know if your drive uses SMR or not.".
Seagate made the first PR move, by getting in touch with ArsTechnica, and swore that, for the time being, their NAS drives do not use SMR (but did not publish a detailed list).
Then Western Digital bucled up under the pressure, and published a full list.
And now, Toshiba published a full list.
So no, not a dupe.
how can it be slower (Score:2)
isn't the read rate exactly equal to the drive rotation speed along the track?
Re: (Score:3)
eh what? the issue with SMR is with random access writing
Re: (Score:2)
Okay same question then. Would not Random read/write be limited by a combination of head seek time and drive rotation speed? Why would the recording modality matter. And it's it's denser then should not all mechaical-factors-being-equal then it be actually faster? I'm wondering what's not "equal" then?
Re:how can it be slower (Score:5, Informative)
SMR writes the tracks in a way that overlaps slightly. That means it can't overwrite a single track, because they would overlap wrong then. This means that to work, the drive needs to first buffer the writes in a special area, then rewrite a part of the disk from scratch.
So, write on a traditional disk: You go to the right track, wait until the right sector is under the head, done. You changed 1 MB, you write 1 MB. Easy.
Write on a SMR disk: You go to the buffer area, wait until the right sector under the head, write. Then, once a bunch of writes add up, you find the corresponding SMR areas, read the whole thing, apply any updates, rewrite the whole area. Here if you changed 1 MB, you need to rewrite a whole 256 MB block (or whatever it is depending on the drive). Perhaps several. Meaning that small changes can potentially turn into very big ones due to how the disk is laid out.
If you write little, this may not matter much, because the drive has enough time in between writes to do all this behind the scenes. But if you operate a busy server and need to write a few TB, then it overwhelms the ability of the drive to buffer, and you're stuck waiting until it goes through all this process.
Re: (Score:2)
Re: (Score:2)
That probably helps, but not necessarily as much as you think. You write a 4K sector to say "hey, the data is over there", but that still goes into a 256MB block that will need to be fully rewritten, and if you need to write those all over the drive, the result is enormous write amplification. You write 16 x 4K blocks, the drive has to rewrite 16 x 256MB blocks in the worst case.
If anything, large writes might be better, because if you're rewriting 256MB to change half of it, much less time is getting waste
Re: how can it be slower (Score:2)
and you're stuck waiting until it goes through all this process.
Or, as others have mentioned, your NAS tells you that the data write action took to long, and it went into sulking mode in the corner..
Re: (Score:2)
So you never read the part about articles how SMR does writing next to an already written track. You need to do that, it is very obvious why it takes a HELL of a lot longer and why it's scary way to do things.
Re:how can it be slower (Score:4, Informative)
Because it has to journal writes in a CMR area before doing a slow commit to the main SMR area. It tries to do the commits during what it thinks is idle time. If you try to initiate a read or write while it's committing the write journal to the main SMR area, the read/write has to wait for the commit to complete. The high latency of the read/write can cause a RAID/storage pool controller to flag the drive as failing.
Re: (Score:2)
thank you for starting to answer my question but it still doesn't explain why SMR is slower. I would expect the rotation speed and track seek times to be unchanged (am I wrong) so the time the head spends over a magnetic domain ought to be no different regardless of the magnetic encoding right? So why would SMR be slower? Is it actually slowing the spindle?
Re: (Score:2)
Re: (Score:2)
They actually are, since drives are pretty damn cheap now.
In fact, the reason so many "desktop drives" are SMR is because of one market - the USB external drive market, which is extremely price sensitive where $10 difference is huge. If you ever scoured sites for the lowest price on USB hard drives, you've done it. Even when they are $10 cheaper.
I've picked up quite a few 5TB portable hard drive
Re: how can it be slower (Score:2)
It isn't slower if the drive is idle. It's only slower if you initiate a read while the drive is committing the write journal (either because it was full or the drive thought it was idle). It's this occasional high read latency that causes the controllers to flag a fault. If the drives read consistently slower that wouldn't cause the problem.
Re: (Score:2)
Re: (Score:1)
No amount of buffering can make the problem go away for sustained random writes. By the time you have enough buffer to cope with random sustained writes, you have so much buffer that you might as well just buy an SSD in the first place.
Re: (Score:2)
That works great until you need to read something that wasn't buffered to the SSD while the SMR drive is in the middle of it's write cycle.
Also, Linus Tech Tips.. really?
Re: (Score:2)
Re: (Score:2)
I honestly don't think most desktops need anything other than SSD. If you get a larger drive, it's because you need the larger space for something in which case, you need to be very careful that your application is capable of handling the crazy write times.
One mentioned case was NAS drives. For me, all of my desktop, laptop, firewalls, etc are SSD only and my NAS is for backups and video files, If my NAS suddenly errors out the drives, that's going to be a huge problem for me. It's workload is exactly the
Re: (Score:2)
Re: (Score:3)
isn't the read rate exactly equal to the drive rotation speed along the track?
Sustained write speed is krap. Is analogous to the write speeds of HDDs of the mid-late 90's on PATA-4. there are tricks to accelerate burst write speed.
Read speed is good unless your read happens to be concurrent with a write or some garbage collection/data redeployment on the drive. In which case, read speed is also Krap.
These SMR drives are not suitable for NAS (even if your NAS is write seldom read a lot) because, as soon as one disk fails and you have to rebuild the array, the slow sustained write sp
Re: (Score:2)
Re:Not a dupe... (Score:4, Interesting)
Anyone capable of running fio will realize something is a amiss.
I have a handful of tests in my back pocket for determining drive characteristics. Maybe handful is a downplay...
They can pull this garbage on consumers until someone blows the dog whistle, but this wouldn't have passed the sniff test in enterprise.
Like others have pointed out, the journal test alone would have nailed these as garbage for anything but cold storage.
This isn't rocker surgery folks.
Granted, I live in a bubble, this shit falls on my desk all the time.
Some days you just need parts and you say 'Hey, this is complete and utter shit, but if you want replacements we can live with it.'
Re: (Score:2)
They can pull this garbage on consumers until someone blows the dog whistle, but this wouldn't have passed the sniff test in enterprise.
Enterprise drives which are SMR are clearly labeled as such in the sticker, the brochure and the spec sheet.
So, in enterprise settings, you do not need those tests to "determine" if the drive is SMR or not. Of course, you do need the tests to qualify the drives, which is best practice.
Also. while in consumer-land all SMR drives are device managed SMR, in enterprise land you also get the option of System Managed SMR drives.
For the Average consumer this IS rocket science. And for many very smart and inteligent slashdot participants, this is also rocket scrience (for instace, computer animators, civil engineers, chemical engineers, you get the drift).
The big problem is not manufacturers using SMR on consumer drives per se. Is the fact that they were using SMR WHITOUT TELLING ANYONE, NEITHER IN THE STICKER, IN THE BOX, IN THE BROCHURE, OR IN THE SPEC SHEET
Re: (Score:2)
I wonder if they are seeing an unusually high rate of returns on those drives. RAID is a lot more popular these days thanks to consumer NAS.
You see stories of people returning a dozen drives, all now used and requiring re-testing and possibly new cases due to unavoidable scrapes while being installed.
Submarining (Score:1)
Why do you keep using that word as a verb?
Re: (Score:2)
It's perfectly cromulent.
Re: (Score:2)
Why do you keep using that word as a verb?
I am not a native english speaker (I tought that my signature of "Suerte a todos y feliz dia!" would be enough of a clue for the smart slashdot readership.).
But, if Chris Mellor, who is not only a native english speaker, but also a journalist (and therefore should have an above average use of the english language) used the word "submarining" a few times in his articles, and even in the headlines, it decided I could use it myself
Namor (Score:2)
Why do you keep using that word [Submarining] as a verb?
Blame Marvel Comics' oldest character Namor "the Sub-Mariner" MacKenzie [wikipedia.org].
ALL HDDs are too slow (Score:1)
I haven't bought a computer with a HDD in several years. Only an SSD will do.
Re: (Score:2)
Re: (Score:2)
Why not?
But no, I back up to the cloud.
Re: (Score:2)
Why not?
But no, I back up to the cloud.
You should back up to two different media in two different geographical locations.
Backing up to the cloud is one method.
Backup up to a USB HDD on your desk is another.
The sweet spot for HDD pricing is for 8TB at about $20/TB.
Re: (Score:3)
Cloud backup services like MS OneDrive or AWS Backup keep copies of your data in multiple geographic locations.
https://www.microsoft.com/en-u... [microsoft.com]
These systems are far better equipped than I am, to manage frequent backups in multiple locations.
Re: (Score:3)
Re: (Score:2)
You should back up to two different media
There's no requirement to backup on different media. Different locations yes. SSD to SSD on the other hand is absolutely not an issue in the risk calculation.
Re: (Score:2)
I'm not sold on the long term effects of SSD storage. Will they be readable after sitting for 10 years? I got a good deal on an LTO-4 tape drive and have been using that for backup.
Re: (Score:2)
Trust me, an HDD won't be working in 10 years either. If you are needing long-term archival storage, you have to get creative.
The key to backups is keeping the storage "alive." If your backup storage is sitting on a shelf somewhere, regardless of the technology, it will degrade. Worse, you won't know _when_ it starts to degrade. The storage has to be kept online, and it has to be repeatedly and regularly verified. Otherwise, your decade-old HDD may be sitting there humming, but have mostly bad sectors, and
Re: (Score:3)
Trust me, an HDD won't be working in 10 years either. If you are needing long-term archival storage, you have to get creative.
HDD's are far more durable then you'd think, far more then magnetic tape which is the common long-term archival storage medium. A few years ago I was doing a switch over at a government office which had stacks of ST3160815AS drives(seagate 160GB drives) mfg '07. When checking them 11 years later they were still perfectly fine. Their magnetic tapes on the other hand had started degrading after 2 years. But really you don't need to get creative with long-term storage. You only need to go with WORM's, and
Re: (Score:2)
Re: (Score:2)
25 years? In 1995, hard drive sizes were measured in megabytes. I don't suppose you have a whole lot to back up on those old drives!
In those days, new dries shipped with a list of bad sectors, that had to be marked off-limits during formatting.
Re: (Score:2)
In 1995, 1GB drives had already been launched to market by other manufacturers. In 1993 Seagate had already started selling 2GB drives to consumer and businesses.
Re: (Score:2)
All right, so you have some 1 GB hard drives that are still running after 25 years! Good for you! I hope your 1 GB of backed up data is really important.
Re: (Score:2)
Well if your assumptions are like your reasoning, I can see why you kinda screwed up...several times.
Re: (Score:2)
"Backing up to the cloud is one method."
Uh ... Cloud? One large cloud provider has three different storage options, which represent SSD, HDD, and Tape. In this case, 'cloud' really obfuscates the reality behind the scenes.
Re: (Score:2)
You can pick whatever medium you like. The point is, using storage managed by someone else.
Amazon Backup, MS OneDrive, Google Drive, and CrashPlan are some good choices. These services keep multiple copies of your data in different geographic regions.
To be sure, there are bad choices for online backup. Do your homework.
Still, a robust cloud backup system is far more likely to stay up and running than your NAS in the back closet. I've had quite a few of those suddenly die.
Re: (Score:1)
Why not?
But no, I back up to the cloud.
There is no cloud. It's just someone else's computer.
Re: (Score:2)
Exactly. Someone else's computer, that specializes in managing storage. Amazon Backup, Google Drive, MS OneDrive, CrashPlan--these are good choices that store your data in multiple geographic locations.
The benefit of using _someone else's_ computer, is that somebody is being paid to keep your data safe. If you choose a good provider, your data is a lot less likely to disappear than if you have a NAS in the closet. I've had more than a few of those suddenly die.
Re: (Score:2)
> But no, I back up to the cloud.
This means you're paying somebody else to back up to hard drive for you. The Internet ambiguates. They use SSD's or nvme for cache. Nobody can afford to keep huge amounts of data on SSD.
Re: (Score:2)
Yep, and I'm fine with whatever technology I'm paying them to use, since they are replicating the data across data centers and geographic locations.
Most of us don't have terabytes of data to back up.
Re: (Score:2)
Speed isn't as big an issue for me- $/TB is more. I was about to buy a Toshiba L200 2TB drive. Glad I didn't. This news came out just in time.
Any recommendation for SSD brand?
Re: (Score:2)
I like Samsung SSDs, but I'd consider Western Digital or most any "brand name" drive.
Re: (Score:2)
A 2TB SSD is nice and all, but at $200+, I just dont see the use case for me to justify that.
For the same $200, I can get
512GB SSD ($60+) and a
4TB HD ($120+)
As for SSD brand, I've not had a problem with my Samsung 840 Pro, which has been going for about 8 years now I guess, 18 TB written to it over the years and S.M.A.R.T. wear leveling suggests I still have 88% of its original write capacity.
I dont need movies/v
Re: (Score:2)
Thank you and Tony too.
A good friend and his fiancee are fairly serious photographers, plus video recording, and go through SSDs like, well, a lot. He said a lot of Samsungs have failed. He's tried many of most brands, and only Sandisk has never failed.
I know Samsung is popular, rated good, and I generally like their stuff, but I worry about data loss. I'd love drives that I can remove the data from the controller. There exist some drives that are a controller with microSD slots, so you have some chance
Re: (Score:2)
Correction: that microSD to SATA adapter gets very poor performance reviews, and looks like it's not available. Seemed like a good idea anyway.
Re: (Score:2)
microSD cards have awful endurance. Run something like a UniFi controller on a Raspberry Pi and it will eat the SD card in no time at all.
Normal SSDs have better endurance, though TLCs are not that great, but can still do the job. a microSD card? Well, they work OK in portable stuff, but even then, something like a dashcam may wear the cards out fast.
Re: (Score:2)
Very excellent points. Is it because they don't use "wear leveling" algorithms for microSD writing?
Re: (Score:2)
Wear leveling in SD cards is not mandatory, but not forbiden either.
Since wear leveling adds complexity to the SD controller, it increases the price, and requires a team that knows what they are doing, soooo..... .... Some of the most reputable manufacturers do it, some of the least reputable manufacturers don't
Re: (Score:2)
Even if they do, the flash memory is probably rated only for a few rewrites. Good enough for a normal camera or a phone, not so good for a dashcam or a frequently updated database.
WD and Kingston have special "dashcam-rated" cards, though at least the Kingston one was eaten just as quickly by UniFi.
In the past cards lasted longer, but I guess they are now QLC and probably low quality QLC at that. That is also reflected in the price.
Re: (Score:2)
Re:ALL HDDs are too slow (Score:4, Insightful)
Re: (Score:1)
Re: (Score:2)
I haven't bought a computer with a HDD in several years. Only an SSD will do.
Congratulations on not having data-storage requirements. I don't have $3200 to spend on storage in my PC.
Re: (Score:3)
Personally, I have ~50TB of usable storage (around 80 raw) in my latest storag
Re: (Score:2)
You're right, I don't need that kind of storage. 95% of us don't, even we who are developers.
For large-scale backups, speed isn't the primary concern. For your primary OS drive, it is.
I worry Tosh's use of SMR on Surveillance HDDs (Score:3)
By its very nature, surveillance drives are write always read seldom drives. That is to say, you are writing all the time, and read only if/when something important happens. That is anathema to an SMR, where sustained writes are slow, reads proceed at normal speed (if those are not concurrent with a write, or some internal HDD housekeeping). As a matter of fact, since the drive is writing all the time, there are no pauses for internal housekeeping.
Unless toshiba is doing domething weird in the firmware, these will not cut it for the intended usage.
The surveillance drives where Toshiba is using SMR are the DT02-V 6TB and DT02-V 4TB. Those are rated for 32 simultaneous cameras. Unless those cameras are motion sensor activaded, not triggered very much, and crappy resolution, I guess that a sustained write rate of a SMR will not cut it.
But kudos to Toshiba (and Seagate) for NOT using SMR on NAS drives (for the time being).
Re:I worry Tosh's use of SMR on Surveillance HDDs (Score:5, Informative)
So in large file write-always application like camera video, as long as you're not filling up the drive (there are lots of SMR blocks free and available), the SMR doesn't really impact write speeds. The video comes in, and gets written in SMR mode in several SMR blocks at full speed. When the security camera switches to writing a different file, the write starts with a new SMR block (or resumes with the SMR block where the previous write to that file left off), and there's no SMR overlap between the two files. You can erase/overwrite one without affecting any other files. It's primarily small file and random writes - where multiple files are written to the same SMR block - which require entire blocks to be erased and rewritten if you change data to one of the files.
That said, this is all how it's supposed to work in theory. I bought one of the early SMR drives and stuck it in my security camera computer. After about a month (when the drive was about 70% full), the security camera program started to report drive timeout errors. I returned the drive and replaced it with a conventional drive, and haven't experienced any problems since. Was this just teething problems when SMR was first introduced, and has been fixed with newer SMR drives? Is there another problem with SMR which is defeating the block writing strategy? I dunno.
Re: (Score:1)
Re: (Score:1)
and yet filesystems are implemented on top of them, and in this case within a fixed function device that can be aware of the underlying SMR technology. Yes, that is how it works, at least how it can work.
Also note that the OP said "this is all how it's supposed to work in theory." The OP fully understands the issues, you.(and williamyf who keeps posting misinformation) do not.
Re: (Score:1)
Re: (Score:1)
That is anathema to an SMR
No it's not the anathema since surveillance primarily requires large storage at a low price. It does not require high performance. It requires good enough performance. You're so focused on technology that you never bothered to ask if the drive is capable of doing what is advertised, which ultimately is sustain writes fast enough to record video. That's not difficult. You're not processing AI data, doing database searches, or anything like that.
As a matter of fact, since the drive is writing all the time, there are no pauses for internal housekeeping.
Drives don't spin down to write at the speed at which data comes
Re: (Score:1)
While all of this is true, it's not the half of it. The OP assumes characteristics of the drives that he doesn't know and assumes that the devices, which are dedicated, fixed function devices tailored to performing these functions, do not know anything about SMR. There is no foundation to support anything he claims is true other than his own biases.
Re: (Score:2)
That is anathema to an SMR
No it's not the anathema since surveillance primarily requires large storage at a low price. It does not require high performance. It requires good enough performance. You're so focused on technology that you never bothered to ask if the drive is capable of doing what is advertised, which ultimately is sustain writes fast enough to record video. That's not difficult. You're not processing AI data, doing database searches, or anything like that.
As a matter of fact, since the drive is writing all the time, there are no pauses for internal housekeeping.
Drives don't spin down to write at the speed at which data comes in. They don't move slower just because you're getting 1.7MB/s off the internet rather than writing 170MB from a local SSD. Drives recording video are most definitely not "writing all the time" unless that video is uncompressed cache from your editing software.
Those are rated for 32 simultaneous cameras. Unless those cameras are motion sensor activaded, not triggered very much, and crappy resolution, I guess that a sustained write rate of a SMR will not cut it.
There's no need to guess. The drives have a performance rating. What I think you're "guessing" is that surveillance is some 4k 60fps deal. It's not. Surveillance recording hasn't changed much in many years and you could handle 32 simultaneous cameras with HDDs from 20 years ago as well, which were slower than these SMR drives are now.
Toshiba themselves say that these drives are supposed to handle 32 cameras at the same time (I saw the specs in toshiba's site).
Toshhiba themselves say that these surveillance drives can be used in NAS as well.
A camera at 1080p (very common nowadays) and doing MPEG4 compression on the camera will produce 1,5Mbps. and that is your best case. Worst case is the camera doing MPEG2 compression only.
And these SMR drives are supposed to handle 32 of those at the same time?
Their only hope is that the cameras are mo
Re: (Score:2)
Why not?
An SMR drive can get around 50-60MB/sec writes continuous. that's around 1.5MB/sec per stream, or around 9Mb/s, which is plenty for even high quality 1080p streams (that is around half of Blu-Ray (video only), double that of any HD streaming se
Re: (Score:2)
Even cheap surveillance cams are capable of doing H.265/HEVC nowadays. DVR/NVR and cams manufacturers usually use their own proprietary H.265 addon these days. So the bitrate will be much smaller. For example in my experience, a camera doing 1080p 15fps H.265+, it only produces ~ 400Kbps (I set 1400Kbps as max bitrate).
Re: (Score:1)
You are wrong because you have only a superficial understanding what happens and what is required of the application. Typical of /. posters of course.
Not all writes are the same, not all write workloads are the same, not all applications have the same constraints, and not all write-heavy workloads are "anathema to an SMR".
Perhaps you should invest your time on gaining a better understanding of the technologies you present yourself as expert in.
Also 6TB and under (Score:2)
As with the Western Digital drives, all the drives that may have SMR are 6TB and under.
So it seems like generally if you want to be safe, get an 8TB drive or larger... not sure why you wouldn't anyway at this point. 8-12 TB seemed to be the best price level of drives last I looked a few months ago.
Re: (Score:2)
Because I only need about 8TB in my home NAS. Currently I'm using 4x4TB in RAID6 (I originally built it using old desktop drives - 2x2TB and 2x4TB, then replaced the 2TB drives). I recently lost a drive and replaced it with WD RED NAS 4TB which was CMR.
I'm intending to eventually replace it with an all SSD NAS - I run VMs and various other services on there that would benefit. In the meantime I'm hoping not to lose another drive as it appears I'd either have to shell out for an 8TB drive where I wouldn't be
Re: (Score:2)
Because I only need about 8TB in my home NAS. Currently I'm using 4x4TB in RAID6 (I originally built it using old desktop drives - 2x2TB and 2x4TB, then replaced the 2TB drives). I recently lost a drive and replaced it with WD RED NAS 4TB which was CMR.
I'm intending to eventually replace it with an all SSD NAS - I run VMs and various other services on there that would benefit. In the meantime I'm hoping not to lose another drive as it appears I'd either have to shell out for an 8TB drive where I wouldn't be using half the capacity, or put in an SMR drive which would severely impact the performance.
You are right in not using SMR for your NAS. Other piece of advice:
If you are using a Synology, and only have 4 bays, a better alternative would be to get 2 SSDs in raid 1 for read/write caching* . And then use the other two bays for a RAID 1 of HDDs. This is the best of both worlds.
If you are using any other NAS, disregard this advice.
If you go all flash, your speed will be limited by both the SATA speed limit, and the 1Gbps Eth speed limit. Besides, synology is known for having a mixture of SATA 2 and SAT
Re: (Score:1)
That sounds like SuperKendall logic. Better yet, only buy from Apple because they will always be superior, and buy your Apple product from a Trump business for that gold-plated experience.
If you want to "be safe" then first understand what "safe" means in your context. Did they not teach you that at your community college, SuperKendall? How can we work an emacs name-drop into this?
Wondering if Seagate SSHD drives would be better (Score:2)
Re: (Score:1)
You are an idiot. A NAND cache will "die in a matter of days" so use an "SSD cache device" instead? You got all the buzzwords in, but none of the requisite understanding. ZFS definitely required as part of everyone's solution!
Re: (Score:2)
Would Seagate's SSHD drives be more responsive with SMR disk technology? On the theory that writes would be cached to the flash and so not cause a shingled write on every metadata update? Sounds like a good combo to overcome the limitations of SMR in this regard.
In principle it could be used. The original incarnation of SSHDs from seagate used 8GB of flash as a read cache and an "inteligent" algorithm to decide what to put on the cache. no write cache.
Nothing stops seagate to put a bigger NAND flash (say, 32GB), and use 16GB for read caching, 8GB for Write caching and 8GB for wear levelling.
Problem is, now we are dealing with a 3 level cache hierachy (the 256MB of ram cache on the drive + 32GB of NAND + the CMR cache area on the drive), the controller also had to d
Unpossible! (Score:2)
Slashdot insisted that WD and Seagate were the scum and that glorious Toshiba would never do this!
Shouldn't have to go check a list (Score:2)
My question is will they change the online descriptions? If they don't, then publishing a list doesn't mean ****.
Re: (Score:2)
I just checked amazon.com for "smr drive" and found several drives labeled SMR but for much higher prices than regular or unlabeled SMR drives. Very odd marketing.