The future of HDDs and Memory

[Runamile]

The thread about the hyperdrive got me thinking. In the far future (think 15-25 years), there is a good chance for many HDDs being solid state. What do you think the plausibility for merging the HHDs with the system memory? It would have to be non-destructive, so it cant be lost then the power is cut(like the great-grandchild of MRAM or something). I would imagine there still being a need for temp space to hold changing data and variables, so that could be put on a separate section of the drive. The 'storage' section and the 'memory' section of the drive can dynamicly change, adapting to the needs of the user.

Say you have a 100TB 'drive' of solid state memory. The user has 75TB of data, to the remaing 25TB is automaticly allocated to system memory. This will shrink or grow accoringly. The tranfers between the two will be very fast, cause it will all be on the same board, and possibly the same silicon. This also means instant shut-down and start-up, since the memory is already loaded. And if one wants to 'reboot', the computer will know it has to wipe the 'memory' part of the drive, not the 'storage' part.

How does this all sound to you guys?

 

[klaviernista]

I work at the patent and trademark office and examine applications relating to magnetic recording media. I can tell you that it is a near certainty that solid state memory is going to be a viable solution in the future of storage drives. Most of the stuff I've been getting is about making it faster, cheaper better etc... The reason is because the recording density of magentic media is finite and the theoretical storage limit for magentic recording media is rapidly being approached.

The density of magnetic recording media has generally been increased by decreasing the grain size of the magnetic layers (through the use of crystollographic structre controlling underlayers) and by increasing the coercivity of those grains (by controlling the composition of the metal alloy used to make the magnetic layer). This has a number of problems:

1) reducing grain size puts more magnetic grains in closer proximity to one another, resulting in cross talk between grains, and hence reduced signal to noise ratio

2) The magentic grains can only be reduced to a certain degree before the grains hit the superparamagnetic limit (basically the grain becomes too small to hold a oriented domain and goes from ferromagentic to superparamagentic). Superparamagnetic materials are useful, but not for magnetic storage.

3) Increases in coercivity of the magnetic alloy are necessary when the grains are small, because the smaller the grain, the more susceptible the magentic domain contained by the grain is to external forces, i.e. heat and magnetic fields. Increasing the coercivity is typically done by adding elements such as Pt, B, or Ru to the Co based magentic layer. Currently, Pt and Ru are the most sucessful, but are extremely expensive due to their rarity.

Thus, it is evidnent from the above that the storage density of magentic disks is finite. Solid state devices work on substantially different principals then traditional magnetic media and so many of the above problems are not an issue in solid state stuff.

That said, solid state memory is widely used currently in the form of storage drives, i.e flash drives, smartmedia, SD cards, etc....

As for using solid state memory as system memory, this is a good idea I think, but current technology is way to slow to be used as conventional RAM. While the technology is sure to be come faster over time, CPU processing speed is increasing every year as well. Furthermore, bus speeds are increasing as well. IMO I think that solid state system memory would be a huge bottleneck in the system because it is quite slow as compared to conventional ram. A quantum leap in technology would be needed for solid state memory to catch up with bus speeds and processor speeds.

However, SS memory is a great option for storage drives and is recording density is growing every year. I just read somewhere about a digital video camera that records DVD quality video on 4GB SD cards.

 

[zephyrprime]

We'd probably need some sort of self-assembly technology to create a good cheap solid state disk because technologies that rely on photolithography produce products that are too expensive to use for storage.