Slashdot Mirror
No Cheap Replacement For Hard Disks Before 2020
siddesu writes with disappointing news to anyone who'd like to see solid-state storage dominate in the near-term future. "A new study of storage technology by the former CTO of Seagate predicts that hard disks will remain the cheapest storage technology in the next decade and probably beyond."
So these people can predict the future now?! Really, you never know what is going to happen for sure. Look at current HDD tech, IBM made the GMR breakthrough and BAM! Huge storage capacity in drives. What makes people think that there cannot be another such discovery with solid state or some other yet unknown tech?
"Some books contain the machinery required to create and sustain universes."-Tycho
...10 TB drives will be $10? More likely, 100TB drives will be $100 but you won't be able to get anything smaller. And they'll still crap out after a couple of years.
Warning: this article may contain humor, sarcasm, parody, and perhaps even irony. Read at your own risk.
If you mean 'lowest cost per GB' then you're probably right. If you mean 'lowest cost per IOPS' then you're already wrong. And if we're talking 'lowest cost for something of adequate capacity and a low power consumption for a laptop' then you're also probably wrong too. When flash drives drop below about $1/GB (and it's already close) there will be little advantage in mechanical disks for most users. It doesn't matter if the disk is bigger if you're only using 10% of the capacity, and it's slower than the alternative and uses more power.
I am TheRaven on Soylent News
The fact a major, respected, industry leader has predicted an upstart new technology will not surpass the incumbent technology is an indication it is almost certainly false.
..don't panic
today, a typical 500 GB hard drive costs about $100
This article must be several years old. In present day, a 1TB hard drive costs about $80.
October 23rd, 2009 By Lisa Zyga
Doh!
"more predictable lifespan"? I take it someone hasn't had a drive head crash... SSDs have wear-leveling and usually on sectors that go bad it is still readable, you just can't write to it. HDDs are more prone to cataclysmic failure compared to SSDs, a SSD usually won't break unless you manage to physically break the circuit board, compared to the fragile platters of the HDD, etc. SSDs fail nicely, HDDs do not.
Taxation is legalized theft, no more, no less.
I mean, both Intel and OCZ have said that once they get to tri and quad-state MLC flash technology, prices should drop considerably by 2012. I think Seagate just doesn't want to be relegated as a dying tech company.
Nobody with a clue has been arguing that SSD's would be cheaper per gigabyte than ye olde spinning-platter HDDs any time soon.
What we're seeing now, and will see much more of, is the hybrid approach of combining a small-ish (80GB) SSD for the most-accessed OS & Apps, with a monsterously huge and relatively slow (array of) HDDs for bulk data archival and backup.
With HDD I/O still the single biggest bottleneck today, it makes sense to start transitioning to SSDs, but it doesn't have to be all at once. The premium for SSDs -- ~$2.50/GB SSD vs ~$0.10/GB HDD -- isn't that much, but it will probably pay for most to wait another year not just for prices to fall more, but for all SSDs to finally support TRIM, and have efficient firmware that competes with indilinx and intel's. SATA3 will also be welcome as current SSDs have already hit the SATA2 xfer limit.
(Oh, and please don't eat the "ZOMG SSDs have limited write-cycles!" FUD. In the vast majority of normal usage patterns, you'll never ever get close to hitting it, and even you did, the failure mode still allows you to READ your data off if you had no backup, as opposed to a HDD crash.)
Power to the Peaceful
Holy shit! Anonymous Coward says current generation SSD's might not be able to be used for long-term backup! Clearly this means there is absolutely no use for Solid State Disks whatsoever!
The game is over gentleman. Time to shut down production and throw in the towel.
Thanks for saving our ass. Anybody reading this should know that Anonymous Coward is a smart cookie--he was also responsible for warning about the transition from perfectly sound MFM interface to the bloated, evil IDE interface. Whatever advice he gives regarding these untested, unsound Solid State Disks should be wisely heeded.
You missed one very important point. Anyone so closely connected to Seagate as a former CTO of Seagate, is very likely to still have connections with that company and that part of the industry. Seagate and other hard drive companies don't want solid state to dominate, as their business is in conventional hard drives. Now we have solid state drives that means the chip manufacturers are now direct competitors with hard drive companies, which means a lot more competitors for hard drive companies.
So any press release has to be considered very suspect at best. Seagate and other hard drive companies would totally loose out if solid state dominates the market so they are never going to admit they are in trouble or that they are in danger of becoming obsolite. No company would admit that.
You're correct, but I have two nitpicks.
First, don't confuse NAND used in SD cards and stuff with the same NAND used in SSDs. They're quite different qualities.
I've had SD cards drop 1/8th their capacity after days of heavy use. SSDs, however, have higher quality NAND and wear-levelling controllers. For Linux, they have better filesystems, too.
Second, up until the newest generation, most SSDs were susceptible to debilitating speed loss after some usage. To be safe, you had to half the benchmarked results. With TRIM and smarter controllers, this is mostly solved, but very heavy usage for extended periods will still result in speed loss. Remember, on an HDD deleting is basically a free operation, but on an SSD it has to physically erase. This would be most noticeable for say... a security camera box recording a dozen or more streams 24/7.
However, SSDs are already replacing HDDs on netbooks
Funny i've noticed things the other way round, all the early netbooks were SSD based but now lots of them have moved to a slightly larger form factor accomodating a hard drive.
note: i'm known as plugwash most places but i screwd up registering that here somehow in the past and now can't register
I wouldn't exactly call PCRAM a breakthrough -- people have been working on it since the 1960s. There are a host of new non-volatile memory technologies that claim to be read for prime time Real Soon Now. Just look at the list of upcoming non-volatile memory technologies in the right column at the wikipedia article. You can't go out and buy most of them yet, but any of them could be winners (besides PCRAM, MRAM is available from Everspin/freescale -- you can buy some on DigiKey if you want).
I wouldn't hold my breath for any of these to replace hard drives, they've been a long time in coming and they're still not really here yet. Hopefully by 2020.
FAIL. You should not swap, period.
Oh, they're so cute when they're young and idealistic! Back in reality, I have a database server with 8 cores, 16GB of RAM, and 500+GB of RAID-10 storage. For all but an hour a month, that's abundantly sufficient for everything we ask of it. For that one hour, though, a bit of that RAID turns into swap while we run some gigantic monthly financial queries.
Your ideal solution would be to spend a few thousand dollars in programmer time to make those queries run faster, or drop at least a thousand on a set of 4GB ECC DIMMs. My practical solution involves allocating 16GB out of 500 to swap for the one hour out of 720 that our normal resources aren't sufficient. Frankly, I like my idea better, and I know that my boss does too.
Dewey, what part of this looks like authorities should be involved?