Submitted by “Toshiba said Wednesday that it had made a breakthrough in hard disk design that will allow hard drives to have much higher capacities than what is currently possible today. The research is in something called bit-patterned media, a magnetic storage technology. The recording surface is broken up into tiny magnetic bits, each of which can hold a single bit of data. The bits are made up of several grains, which are organized in rows. This organization is what makes it possible for data to be found easily.”
…on the site the article originated from concerning transfer speeds (disk to disk, backups). It is time to develop a faster bus of some sort. I think memory buses and the interface to the CPU has improved, certainly PCI x16 is nice but there must be a better way to transfer data en masse.
you mean like a fiber channel bus
That’s the ticket!
Intel has on-chip lasers now. I think that will be the future and soon motherboard traces will be obsolete as chips beam the data to each other over light.
Could be a problem if the inside of your case gets really dusty.
fujitsu already had a working prototype in 2007
http://www.reghardware.com/2007/08/13/fujitsu_hdd_update/
… but the prefix ‘mega’ implies more than a fivefold increase in storage capacity.
Hi,
Um, no. Modern drives are “Tera” (e.g. capable of storing 1 TiB or more), so the prefix “Mega” implies a 1000000 times reduction in storage capacity…
– Brendan
Edited 2010-08-22 04:47 UTC
I would gladly pay five times the price of a current 1 TB hdd for a 80 GB disk which transfers data at 40 MB/s, accesses it in ~60µs in the common case, and is guaranteed to store it reliably for 30 years.
What’s the point of higher-density hard drives if they suck just as badly as current hard drives from access times and data storage lifetime points of view ?
Edited 2010-08-21 09:54 UTC
I don’t know what disks you are using but the majority of hard disks sold these days can easily achieve 40MB/s without any problems – the biggest barrier right now isn’t throughput but latency which is related to the fact that we rely on spinning disks. Until there is a technology that is fast, cheap and provides plentiful storage you’ll always be choosing between an expensive flash based solution or a cheaper and slightly slower hard disk solution that has heaps of space.
For me right now I have a 640GB Samsung hard disk that is very fast, very reliable, low latency and great power management – all working perfectly in my MacBook Pro. The hard disk maybe old fashioned but it does the job it was designed for incredibly well.
You can overcome the spinning disk barrier if you are ready to lose storage density. The larger the surface occupied by a bit, the faster you can make the disk spin because it’s easier to locate and read a bit.
Doesn’t the data take up more physical space on the disk platter if the storage density is lower. So if platter is spinning at 7200 RPM, the disks with the higher density will have higher sequential read and write speeds.
This leaves seek times as the major bottleneck with regards to HDD performance.
You would just bring tons of problems:
1) Noise, faster you spin more noise.
2) Heat, faster you spin more heat you generate.
3) Reliability, faster you spin more wearing is caused to parts. Also heat will cause more wear.
4) Cost, you need more rigid parts, more ventilation and insulation to overcome problems 1-3
Solid state drives (SLC) maybe what you are after, however 30 years of testing has not occurred yet.
Yay, pointless anecdotal benchmark follows:
My main PC has 2 drives. 1 SSD, 1 HDD.
SSD: OCZ Agility (64GB)
Average access time: 0.2 ms
Average read rate: 158.3 MB/s
HDD: Samsung Spinpoint (1TB)
Average access time: 13.7 ms
Average read rate: 83.3 MB/s
If you tailor what technology you use then performance is good.
Seriously, time to leave the technology behind. This is like working harder to encode 5.1 surround sound out of a vinyl record. The market needs to progress, and that progression lies in chip-based storage, or whatever 3D optical-based tech that’s hidden away in intel or Bell Labs-based skunk-works waiting for every last marketable penny to be sucked dry from the consuming masses who are purchasing the latest stale, magnetic-spinning based tech that basically hasn’t changed for the past 50 years.
The computing world does not need one 7200 rpm drive to be able to hold 5TB. It needs evolve the storage medium through being the performance bottleneck that it is.
</end soapbox>
Higher capacities, great… but what about higher reliability? The higher the disk, the more data gets lost when the whole disk fails, and the higher the rate of read/write errors. This silent corruption is probably even bigger a problem than total failure.