Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

Disk Prices on Amazon

diskprices.com

Post reply on HN
Print view View original

171–180 of 224 posts

Re: Disk Prices on Amazon

#171

Post by grosswait »

SlowRobotAhead wrote: post #118
concerned_user wrote: post #110

Earlier quoted context omitted.

Software raid (implemented in os, not the fakeraid/softraid chip on motherboards) in this case is better because it can use as much memory as it needs and also since it is part of the OS it can receive additional information on what kind of workload OS wants from it, i.e. readahead or not and how much. Hardware raid generally doesn't know about those things and tries to infer them so it can't optimize fully.

Sort of, I’d have no issue with software RAID. VMWare however has exactly ZERO software raid options so on servers you are limited to HyperV it seems.

Or the myriad of Linux based options

Re: Disk Prices on Amazon

#172

Post by ody4242 »

SlowRobotAhead wrote: post #118
concerned_user wrote: post #110

Earlier quoted context omitted.

Software raid (implemented in os, not the fakeraid/softraid chip on motherboards) in this case is better because it can use as much memory as it needs and also since it is part of the OS it can receive additional information on what kind of workload OS wants from it, i.e. readahead or not and how much. Hardware raid generally doesn't know about those things and tries to infer them so it can't optimize fully.

Sort of, I’d have no issue with software RAID. VMWare however has exactly ZERO software raid options so on servers you are limited to HyperV it seems.

> so on servers you are limited to HyperV you are limited to HyperV, Xen, KVM or insert your favorite hypervisor except for ESXi

Re: Disk Prices on Amazon

#173

Post by londons_explore »

rasz wrote: post #168
londons_explore wrote: post #145

Earlier quoted context omitted.

Nobody puts any R&D into new drive technologies. If demand drops, factories close since they are producing at near zero margin. All R&D costs are already amortized or written off. End result: Static prices. The only time prices will change is when only one manufacturer is left and they jack the prices as a monopoly on antique tech that a few big companies still need. We're a long way from that tho.

>producing at near zero margin :)) https://ycharts.com/companies/STX/gross_profit_margin SSDs are unsuitable for long term storage, technology is simply incapable (cells losing charge, normal usage physically degrading medium, limited number or charges per cell). We dont have anything better than spinning magnetic medium at the moment, and nothing viable on the horizon.

> cells losing charge

Leave it powered up and let the controller re-write the data every year or so automatically.

>normal usage physically degrading medium

For 'read-only' data that is rewritten once per year to compensate for cells loosing charge, you can expect a lifetime over 10,000 years.

Sure, some SSD's available today wont handle the above cases properly and data will be lost, but it is not a theoretically unsolvable issue.

Re: Disk Prices on Amazon

#174

Post by willis936 »

rasz wrote: post #168
londons_explore wrote: post #145

Earlier quoted context omitted.

Nobody puts any R&D into new drive technologies. If demand drops, factories close since they are producing at near zero margin. All R&D costs are already amortized or written off. End result: Static prices. The only time prices will change is when only one manufacturer is left and they jack the prices as a monopoly on antique tech that a few big companies still need. We're a long way from that tho.

>producing at near zero margin :)) https://ycharts.com/companies/STX/gross_profit_margin SSDs are unsuitable for long term storage, technology is simply incapable (cells losing charge, normal usage physically degrading medium, limited number or charges per cell). We dont have anything better than spinning magnetic medium at the moment, and nothing viable on the horizon.

Then we don’t have anything at all. Hard drives are only acceptable at long term storage if they’re unplugged, and even then there is a significant risk that the data won’t be accessible in 50 years. If 5 years is considered long term, then SSDs are acceptable. If 20 years is considered long term, well, make sure you have a robust RAID.

Re: Disk Prices on Amazon

#175

Post by bangboombang »

dehrmann wrote: post #107
sn wrote: post #32

We stopped buying drives from Amazon when we got drives for "the wrong region" and couldn't RMA them. NewEgg is a bit better, but don't buy refurbished drives. Provantage is another reputable vendor that usually has decent prices. If you're local to the SF bay area, check out Central Computers. King Star USA also works if you're a business customer making a large purchase.

I bought some "Amazon Renewed" WD/HGST 10TB He drives. They came from a sketchy seller that provides a "warranty," but the provided phone number and website didn't match. I think one was used but a different(?) sketchy used HDD seller in another state. I'm pretty sure there's no real warranty, I have no idea where these drives even came from, I have no trust in "Amazon Renewed," and I have a fear the drives are actua…

I tried that once a couple months ago. The drive pretty much came with a head crash preinstalled. I plugged it in and it sounded like it's shredding bricks in there.

I assume the seller just sources defective drives from somewhere and directly resells them as refurbished, letting the buyer do the actual testing. I assume this must work because some drives might erroneously been marked defective when it was really a sofstware issue, or other hw problem, like faulty sata cable, mainboard, loose connection, and then obviously there are drives with minor issues, like just a few reallocated sectors, that the naive buyer might not be aware of.

Re: Disk Prices on Amazon

#176

Post by giancarlostoro »

wtallis wrote: post #72
giancarlostoro wrote: post #42

Earlier quoted context omitted.

Hard Disks sure, but I've gotten a reasonable choice of SSD's even one that was faster than any SSD I've owned thus far.

It's really easy for Best Buy to stock Samsung SSDs and therefore offer you something faster than anything else you've ever owned. But that doesn't mean you aren't overpaying by at least $40/TB compared to a good Amazon or Newegg price on an almost-as-fast drive from a smaller brand.

It wasn't a Samsung SSD, I don't really buy Samsung since it seems overpriced and overhyped most of the time, unless I catch a sweet deal. I had bought a Sandisk that was actually more performant than I anticipated, it was noticeable immediately after installing Linux on it.

Re: Disk Prices on Amazon

#178

Post by zaphod12 »

Kaiyou wrote: post #169
cnst wrote: post #71

Earlier quoted context omitted.

I don't understand why anyone still buys anything at Amazon. Their Fulfilled-by-Amazon (FBA) idea was quite a game changer, but the implementation with the inventory co-mingling is something that I'd much rather stay away from as a consumer. There's been oh-so-many reports that even if you're buying from the official store of any known brand, with FBA, then you might as well still receive a knock-off, due to the inve…

Trust. If something happens, I know amazon got my back. Unless they betray my trust, I will continue shopping on amazon.

Funny timing - this s theVerge's headline story: https://www.theverge.com/2020/1/28/21080720/amazon-product-l...

Sometimes waiting too long for your personal trust to be broken (after seeing them drop on others) can be a lot worse than just a broken toy.

Re: Disk Prices on Amazon

#179

Post by close04 »

londons_explore wrote: post #147
dmos62 wrote: post #146

Earlier quoted context omitted.

Audit-based security is ok, but the ideal solution is open hardware and firmware.

My ranking of security: "It's secure because we say so" "It's secure, and we have a published design and audit" "It's open source" "We're offering a reward to anyone who can break it" "Our company would go under if anyone found a flaw in it" SSD's are currently at the first stage.

I'm not sure if I'd put being open source higher than actually being audited. In theory there's no difference between the 2. In practice being open source means that it could be audited while publishing an audit means it was. It's the difference between giving someone money and giving them a lottery ticket for that money.

As many programmers admit themselves, it's also their job/duty/etc. to audit open source code but most actually don't especially for the sprawling, complex code. When everybody's responsible nobody is. It's better than closed source and security through obscurity but potential is only valuable if exploited. OSS without audit could be in some cases just as (in)secure as closed source simply because it may provide its own layer of obscurity through complexity and the expectation that "someone" will check. Someone else.

Re: Disk Prices on Amazon

#180

Post by rasz »

londons_explore wrote: post #173
rasz wrote: post #168

Earlier quoted context omitted.

>producing at near zero margin :)) https://ycharts.com/companies/STX/gross_profit_margin SSDs are unsuitable for long term storage, technology is simply incapable (cells losing charge, normal usage physically degrading medium, limited number or charges per cell). We dont have anything better than spinning magnetic medium at the moment, and nothing viable on the horizon.

> cells losing charge Leave it powered up and let the controller re-write the data every year or so automatically. >normal usage physically degrading medium For 'read-only' data that is rewritten once per year to compensate for cells loosing charge, you can expect a lifetime over 10,000 years. Sure, some SSD's available today wont handle the above cases properly and data will be lost, but it is not a theoretically un…

1 more like every 30 days https://www.anandtech.com/show/8617/samsung-releases-firmwar...

2 No such thing as read only SSD. Just like in ram, SSDs experience read disturbance (rowhammer) https://users.ece.cmu.edu/~omutlu/pub/flash-read-disturb-err...

"A key contributor to this reduced reliability is read disturb, where a read to one row of cells impacts the threshold voltages of unread flash cells in different rows of the same block. Such disturbances may shift the threshold voltages of these unread cells to different logical states than originally programmed, leading to read errors that hurt endurance"

The very best SSDs today promise to go into read only mode in case of a failure, and in tests almost none are. Intel server family of drives famously brick itself despite stating "read only" in documentation.

Post reply on HN