Slashdot Mirror
Hitachi Promises 4-TB Hard Drives By 2011
zhang1983 writes "Hitachi says its researchers have successfully shrunken read heads in hard drives to the range of 30-50 nanometers. This will pave the way for quadrupling today's storage limits to 4 terabytes for desktop computers and 1 terabyte on laptops in 2011." Update: 10/15 10:39 GMT by KD : News.com has put up a writeup and a diagram of Hitachi's CPP-GMR head.
Trying to build an open source PACS system at a hospital I consult with. The need is basically for lots and lots of storage, without needing to access a DVD or tape. A typical MRI / CT scan can generate 1 GB of data; so with dozens of scans a day; and the need to store and access patient data pertaining to say, 10 years; these drives will be really useful.
A simple SATA RAID controller interfaced with 4 such drives can give me 12TB of cheap, fast, storage. At 1TB per year, should be good enough for my needs. H/w vendors currently recommend expensive SAN boxes; which I don't like... no useful value for the application at hand.
If you keep throwing chairs, one day you'll break windows....
With the current market trends, the flash memory-based HDs should be cheap enough to replace magnetic hard drives in laptops by 2011 in most applications. They are already superior in access time, drive life, power use, and transfer speeds (see the FusionIO demo or MTRON drives).
Actually, my first reaction was, "That's all?"
They're talking about having this capacity available in another four years, and yet, 4 TB isn't even that much now. I have four drives in my computer totaling a little over 1 TB, and since the start of the year, it's mostly gone. A few uncompressed videos, a decent music collection, and a handful of the latest games... suddenly you're trying to decide what you need to delete before grabbing the camera and starting a new project.
(My work and hobbies all revolve around video, but I know plenty of people who could already fill that drive with just games, movies and porn.)
This will pave the way for quadrupling today's storage limits to 4 terabytes for desktop computers and 1 terabyte on laptops in 2011.
Prior to the rise of perpendicular recording, we had cheap and plentiful 200-400GB HDDs using plain ol' longitudinal recording. Suddenly PMR hits the market, promising 10x the storage density at up to 1Tb/in^2 (which Seagate claims they actually achieve), and two years later we have only two real models (with a few variations for SATA/PATA) of 1TB drives available.
Call me crazy, but a few really trivial calculations show that at 6.25in^2 *of usable area) per platter surface, times two surfaces per platter, times three platters, we should have, using today's technology, 4.5TB (note the change in case of the "B", no confusing units here) 3.5" HDDs.
So forgive me for not wetting my pants in excitement about an "announcement" that something realistically available today, we won't have for another half of a decade.
The real problem is not the lack of space but the systematic chronical unability of the industry and users alike (but especially the industry) to properly manage their files.
Yes, there are some cases where 4TB truly isn't enough without the problem being poor data management (large datacenter, huge DVD-quality media collection, etc). But far too often we see the reason for more space being poorly managed mail servers, tons of WIP that has not been properly archived or disposed of, huge amounts of unhandled spam, work-related casual conversations that really don't need to be stored after the work they relate to has been completed, outdated and obsolete software not being uninstalled, inflated registry (or any other overhead data) that keeps being backed up and restored without any cleanup involved...
A lot of people, when challenged with the problem of this vast array of useless junk data will just respond "well we have space, and if we run out we can always buy more, and the purchase price is way cheaper than the manhours needed to clean up this mess, so why bother". Another common excuse is "it doesn't bother me, so why not keep it just in the potential case I'll ever need it again, even if the chance is extremely small".
It does not occur to these people that proper data management is extremely important procedure, and must be ingrained in the business process. Much the same way you clean up physical garbage, remove obsolete physical equipment, empty the contents of that blue recycle bin under your desk, and do it all on a regular basis to keep the garbage from getting out of hand. Trash not worth keeping in real life does not become valuable when stored online, even if it can be stored for free or cheaper than the disposal price.
Properly disposing data as a business process will take time, but this time will be saved many times over when people don't have to dig up through junk to find what they need, when important things are not buried in crap, when all data worth storing is clean and polished and free of rust, when your OS is not clobbered up by crap processes or temporary files, when your DBE doesn't have to go through zillions of crap stored in the database to find a single row, when you do the cleanup as-you-go, rather than waiting for things to be completely out of hand and then doing a half-assed job because by that point it is really hard to tell apart the good from the junk.
The problem is spiraling - the longer people don't properly clean up data, the harder it is to clean it, especially as files grow larger and more complex as hardware and applications evolve. In turn, it motivates people to just invest in extra drive space, processing power, memory, etc, because by that time it's cheaper than the cleanup. And of course, once the resources have been invested into, they are filled with even more crap until they are full too.
But the biggest problem of poor data management is actually not technical, it's business-related. As we are faced with an increasing information overload, it is very easy to make poor decisions based on data that is not necessarily wrong, but is outdated, matched with incompatible other data, or just not put in the right perspective. The whole "data warehousing" principle absolutely REQUIRES proper and timely maintenance and cleanup of data. This is so important that (and this has been proven over and over again) large corporations with proper data management gain a substantial strategic advantage over those who don't.
It's not just about a little slower response time, or some more work to find what you need on the server. It's about right business decisions vs. wrong business decisions. And it's also about not being taken advantage of - contractors and business partners can easily manipulate data to present it in the light favorable to them, and if you are a private business, this kind of crap can make you bankrupt. Of course, it happens day after day in the government with the taxpayers footing the bill, but that's another story altog
I want more reliability. Over the last ten years of using hard drives, I have about a 50% failure rate.
I see comments like this all the time, and really don't understand them.
I have personally bought an average of one HDD per six months over the past decade, and, except for ones outright DOA, I have only had one fail, ever (and that after it had served for a good many years). And I include both DiamondMaxes and the legendary DeathStars in that list, both considered some of the most prone-to-failure out there.
Considering my work environment, I can expand that sample to most like 100+ HDDs; Of those, only two have failed, both laptop drives.
I have to suspect the people experiencing the flakyness of HDDs either fail to adequately cool them (I put ALL my HDDs loosely-packed in 5.25 bays with a front-mounted 120mm low-RPM fan cooling them) or somehow subject them to mechanical stresses not intended (car PC? portable gaming rig? screws tight agains the drive's board?).
Sound nuts? Yes... but they do. Large clusters of many inexpensive machines set up in a redundant manner...
I run: Windows, OS X, Linux, FreeBSD. Just because you have a hammer, doesn't mean everything is a nail.
I agree with the stagnation part. At work some of our laptops are more than 4 years old (May 2003) and they are still perfectly capable and working (P4 @ 2.8 GHz, 512Mb RAM, 60GB HDD). We even have two T30 Thinkpads that are just enough when traveling to browse, check email and write a doc.
Regarding the second part (reinstalling XP) - you should really look at Acronis True Image - it's what we use.
Basically, you install WinXP+patches and whatever programs you need once, make an image and store it on a DVD, network or on a hidden partition on HDD. At boot, you can press F11 to start Acronis instead of Windows from the hidden partition (it's a lightweight Linus distro) and you can restore your image in 5-10 minutes. Even if the image is 6 months old, you still need to download just a few patches and software updates (e.g. update from FF 2.0.0.0 to 2.0.0.7).
1 CPP-GMR: As an alternative to existing TMR heads, CPP-GMR head technology has a lower electrical resistance level, due to its reliance on metallic rather than tunneling conductance, and is thus suited to high-speed operation and scaling to small dimensions.
2TMR head: Tunnel Magneto-Resistance head A tunnel magneto-resistance device is composed of a three layer structure of an insulating film sandwiched between ferromagnetic films. The change in current resistance which occurs when the magnetization direction of the upper and lower ferromagnetic layers change (parallel or anti parallel) is known as the TMR effect, and ratio of electrical resistance between the two states is known as the magneto-resistance ratio.
Source: Official Press Release
The important thing is not to stop questioning --Albert Einstein.
Software Raid Sucks.
Bzzzz. Next contestant!
Linux's software RAID solution is often faster or on par with high-end hard RAID solutions, plus, it doesn't tie you to a specific hardware vendor. Linux's software RAID solution is generally far, far, far better than the low-end, commodity RAID solutions which comes with various MB/chipsets these days. The down side of software RAID is it takes more CPU. In a day where multi-core CPUs are common and CPUs are faster than ever, almost everyone can spare the CPU. The combination of the two means software RAID is actually one of the best solutions available for many classes of casual RAID users.
I use a custom file system, that I wrote myself (woot FUSE), that hides duplicates, so that only one real copy exists but the file system behaves as if multiple copies can exist, and compresses files that will have 10% or greater reduction in size. Not that it matters to this discussion, but in case anyone is going to comment, it caches read files in memory so it doesn't have to constantly have to decompress files. It caches file writes on disc and only compresses the files once there have been no further writes for a given amount of time. It also does versioning. I'm working on making it flag rapidly changing files, such as bit torrent downloads, so that they won't be processed to save space or keep versions, until they've finished downloading.
:)
I still need terabytes of disk space.
At what price learning? At what cost wisdom? The price is a man's peace of mind, and the cost is his life.
No, I mean they need to figure out how to fit multiple disks and a RAID-5 controller in the size of a single disk so that it can be installed as a single disk and only worried about if something goes wrong. It needs to be extremely easy to use. Maybe make each mini-disk from the RAID so that it can be ejected and replaced without opening the case and with a little light on the front of the case for each mini-disk so if it turns from green to red you'll know it's broken.
They've squeezed enough space into that size for now - now I'd rather they work on reliability. For most people reliability is more important than hdd size or speed. I'm actually surprised Dell or one of those big name sellers hasn't started pushing the reliability angle as it seems such an easy upsell. "You wouldn't want to lose your family photos or important documents would you?"
Us geeks already know enough to use RAID, and a lot of us do use Linux's RAID 5 support, but they need to make it the default option, and really easy, so that everyone uses it. Even if they could make a multi-hdd unit with built-in RAID 5 that would fit in a 5.25" external bay it'd be useful.
At what price learning? At what cost wisdom? The price is a man's peace of mind, and the cost is his life.
I write auto-sorting programs as a hobby which is part of why I download so much. I used to have the image portion tied into the system so users could browse images, rate them, keyword them, etc but it became sort of pointless once Flickr came out so I stopped pushing my own image indexing and search tool.
:)
I have a toy, which I keep considering turning into a business, that would make it easy for users to backup their files to a central server farm that'd keep redundant copies in different locations, make files easy to restore, share, index, search, etc. It's meant to work with files of all types and is built on top of a file-system I wrote that makes storing files space efficient by removing duplicates and using compression. I personally think it's sort of a killer app since it combines network backup, file sharing (with built-in BT), file tagging, powerful search, live previews, forums, an open API so other web apps can use it's functionality, etc all into one but I don't think it's the kind of thing most financiers would grok. I wish I had a buddy that was into business. For fellow geeks, or people that have at least read Cryptonomicon, an Avi.
At what price learning? At what cost wisdom? The price is a man's peace of mind, and the cost is his life.
--D00D, I like your idea! Make it a self-contained ZFS + Samba system with 3-5 removable "platters" in the form factor of ~2 5.25" HDs stacked on top of each other, and sell it as an eSATA device for Backups! :)
--2-port PCI SATA cards can be had for like $19 these days, and they DO make eSATA -> Sata cables for ~$7.
.
== WolfriderV6 == I'm willing to admit that *I just might* be wrong... Are you??