After quietly switching to slower NAND in an NVMe SSD, Western Digital promises to be a bit louder next time Western Digital says it will alert customers when it reformulates its products by modifying their firmware and electronics, as opposed to burying salient changes on a spec sheet without any public announcement. This issue came up lately when the computer storage giant low-key altered the components in its WD Blue SN550 NVMe …
switching up their firmware and electronics
Who are they kidding? The product's specs are now substantially worse. They must be forbidden from doing this to products without at least changing the model number, not just the spec sheet. As it is now, customers have no way of knowing what specs they're getting when they buy an SN550: don't even think of buying one unless it's cheap and the lower specs meet your needs and expectations.
"The product's specs are now substantially worse."
From what I saw, the performance whilst writing to the SLC cahce (12GB) was the same as the old version, and then it fell of a cliff. Therefore, it does still meet the spec sheet, but means the reviews are now misleading (and no, I'm not saying this is OK - it's not!).
For what it's worth, ADATA (XPG SX8200 Pro) and Crucial (P2) have also done this recently. :(
> For what it's worth, ADATA (XPG SX8200 Pro) and Crucial (P2) have also done this recently. :(
Add Samsung to the list with recently manufactured 970EVO Plus drives.
Oh, this is NOTHING, they're amateurs compared to network interface manufacturers. D-Link, for example, has 4 revisions of just a single *one* of their gigabit ethernet cards, the DGE-530T
https://support.dlink.com/ProductInfo.aspx?m=DGE-530T
and, joy, each one takes a completely different driver. Pick the wrong hardware revision level, download the wrong driver, and (if you are not tech savvy) you'll be pulling your hair out trying to figure out why this thing won't work.
And this level of stuff has been going on for decades. Want a fun read regarding the joy of network adapters from almost 20 years ago?!
https://www.tek-tips.com/viewthread.cfm?qid=288507
FIVE different drivers for the Linksys 100TX card, due to 5 different hardware revisions.
Or, how about chip bugs only solved by finding a completely different manufacturer's driver??
https://www.tek-tips.com/viewthread.cfm?qid=289770
And yes, that's me giving out the tech support.
<blockquote>This is why i would NEVER recommend "hardware RAID" over the fantastic stuff that can be done with the Linux kernel and mdadm.</blockquote>
Your normal SATA ports don't power-down a drive having problems, they just keep trying to read it. If you're lucky, the drive responds back with an error or timeout quickly and you don't notice it too much. If you're less lucky, one bad hard drive makes your whole system hang in your software RAID setup. If you're really unlucky, one hard drive is reporting an error here and there, but still mostly responding, even as it doesn't write all data to disk, which you won't find out until the next scrub/check or the "good" drive failing and needing to be rebuilt from the marginal one.
Hardware RAID is always smart enough to disable a misbehaving drive at the first sign of trouble and never crash your system because of a bad drive.
I also don't see how this relates to the parent post, as LSI/Avago is heavily dominant in this space. So much so that Linux kernel md RAID can read and use LSI MegaRAID signatures, so you can continue to use your hardware RAID volumes without a controller.
LSI/Avago has drivers easily available and doesn't pull any tricks like silently changing their cards because the enterprises that are their primary customers would destroy them for it.
Problems you are/have had with "hardware RAID" were more likely proprietary software RAID, the options for which are commonly included in the BIOS/EFI firmware of higher-end power-user motherboards, but really don't do anything without their driver loaded, which does the software RAID bits.
Similar circumstances arise when sourcing video capture and/or TV tuner cards. Windows users are usually supported by the manufactures drivers, but we Linux and FreeBSD users often have no idea which chipset is on the card, even if buying an identical model to a known working one.
I was shocked to find that their years-old factory firmware corruption that makes their hard drives instantly spin down is still a thing. If I buy 5 drives for a RAID and 3 are in an endless spin-up-down-up cycle, I'm not going to keep returning drives until I get a good set. They're all going back for a refund and I'm buying another brand.
Which brand? WD has the majority of the spinning-rust market. Seagate is about half the size (and pulls lots of nasty tricks of its own), and all others are noise. Toshiba doesn't seem to be trying very hard. The rest are gone.
There isn't much more competition in the enterprise SSD market after WD purchased HGST and Sandisk, just leaves them, Samsung, and Intel as major players there. Quite a few hungry to get in the game, with little success. The consumer SSD market is more diverse with tiny players trying to get a foothold.
An increasingly desperate company that reported earnings of almost $5 billion for Q4 2021 results.
I'd like to be desperate like that.
Their gross margin is 26% though, which is huge!
you probably haven't done a LOT of financial analysis and reporting for the top-level executive suits, have you?
When it comes to cost accounting and analysis, there are those factors (generally called called "burden") that are basically fudge factors applied to adjust estimated costs for specific product lines in order to determine what it REALLY costs as best as possible without getting down to the nit in every analysis.
"Gross Margin" generally takes some (but not all) of this into account. As a general rule, the gross margin for a product line is used to determine "which products are profitable" so they know what to can, what to cost reduce, and what to promote like hell. That sort of thing. But it's "gross margin" and STILL does not factor in ALL of the costs. (Obviously that comes later with the 'bottom line').
Not unexpectedly the costs of sales, marketing, "ivory tower", taxes, loan interest, dividends, and other such things take a HUGE bite. That's where all the margin disappears to, as a general rule.
So 26% gross margin in a financial report might simply be fudged up or down to look better than it is, or may depend heavily on how they do their overall cost accounting at the end.
/me takes off accountant hat and tosses it back into the corner - stupid stinky hat
Do you know what “gross margin” means? It means the difference between the cost of building and selling one drive, compared to the revenue. Gross margin has to pay for research and development, for building factories, for advertising and running stores, basically for all the fixed cost. You can have 24% gross margin and lose money, because your development cost too much.
I don't just see WD doing this sort of thing, its been going on in the tech industry for years, change the specs after the products sold a bucket load and just quietly update the spec info without changing the model number. That way you can ride hide on tech reviews for the early units using higher quality components, as how many reviews are done of a product 6 months after its been on the market?
I remember Kingston did this with one of their network cards back in the early 2000s, we bought a load of Ethernet adapter which came with one chipset, and then after a few months they switched to ones with another chipset without updating the model number and when bought 100s more to find they were incompatible with the desktop image we had created with the first cards drivers.
While switching the chipset isn't ideal (for an imaged environment), Kingston should at least have used a different model identifier... i.e. /a, /b, or so on.
However that is different to what many SSD manufacturers are doing now which is nothing other than scumbag bait-and-switch - release a product, get good reviews, replace the product with something cheaper and inferior.
I get replacing chips and so on, after a product release, and particularly these days with the supply issues - but replacing chips and supplying a product that is inferior to the original is dishonest.
Some laptop manufacturers also keep the same model number while seemingly changing the components weekly.
Oh like the Dlink DGE-530T where revision A and B were 3Com/Marvell Yukon chips and then revision C was a DGE-528T (which no one wanted) using a realtek 8169. Far inferior network card and of course totally incompatible with drivers, and since dlink didn't want people to know it was a realtek it had it's own PCI ID so linux didn't even know it was a supported chip until someone (me in this case) added the PCI ID for it to the kernel. We of course told IT to stop buying those cards anymore.
Virtually every company tech and non tech is guilty of doing this.
The spare parts you buy to find out that they only fit a model made up until X date provided the serial number is less than something.
Just try and buy spares for a car without the chassis or registration number you will find a subtle difference where the replacement won’t fit.
I had a bmw where one year into the production run they changed the door lock mountings as a result you need to check the part number against the chassis or it won’t fit. But look at two with the different doors and nobody can spot the difference as it is all internal.
It normally - on OHV engines at least - sits between the turbo encabulator and the muffler bearings. If you’ve ever experienced your car failing to start on the first twist of the key, or jolting slightly as it up/downshifts, or seeming to use slightly more/less fuel than it used to, then this small but vital component should be your first suspect.
Trust me, am totally serious mekanik.
"they switched to ones with another chipset without updating the model number and when bought 100s more to find they were incompatible with the desktop image we had created with the first cards drivers."
We had the same with a model of Intel motherboard. No model# change but a whole other chipset for the onboard NIC and had to re-do the standard deployment image.
This one, the D915GHA
From that spec, it seems the two different NIC options were factory build options. I don't recall that at the time, nor any difference in model numbers. I suspect what happened is they found most people wanted the Gb Marvel NIC rather than the 10/100 Intel NIC and simply abandoned the 10/100 Intel option in favour of the Marvel Gb option.
When I used to build pc's the goalposts changed constantly. A better component would be offered at more of less the same price so I would get that instead, sometimes the usual component would be not available or increase in price. Hard drive capacities would be a good example of this, all other things being equal. There are however examples where bigger/better can cause problems though - with Hard Drives there can be BIOS gotchas, if not careful.
Sucks too. I (among many others, I'm sure) was stung hard by the Seagate 3TB drive debacle, and I've been recommending/buying WD drives ever since.
But between this and the Red SMR fiasco, I just can't trust Western Dij anymore - whoever's in charge there is clearly putting this-quarter profits ahead of all the goodwill they've built over the last couple of decades. Did they get bought by a hedge fund or something?
I'm very happy with the Toshiba drives I've purchased, but they're apparently out of the game now. Is there anyone reliable left?
I was wondering the same thing. Seems that every storage manufacturer has either gone full scumbag, got bought out by one of the former or pulled out of the market. Back in the day it was much easier, i.e. "avoid Seagate" in my case (I got stung by a large number of their 1GB drives failing; fortunately backups etc but it was pretty annoying) but these days they all seem to be problematic.
Tough but fair...
The weird thing is that these various outfits must be selling container loads of storage to "big cloud" who I'm sure wouldn't tolerate this kind of behaviour for an instant.
That implies they're going out of their way to create crap for the plebs. Question is whether it's "because they can" or because the cloud mongers are squeezing their margins to the point that this is their last remaining source of profit.
I suspect they are screwing over the plebs because component shortages mean they are putting the reduced quantities of the good stuff in the enterprise kit. There's little to no good stuff left for the retail[1] drives.
[1] Probably applies to stuff sold to the box builders too.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
