Quote:
Originally Posted by Multiverse
Moving parts with all its drawbacks - no thanks, don't need 'em. This technology for storage will come to a dead-end. Today's Solid State Drives are just the beginning of our future of data storage.
|
Hi Grant,
Thanks for the post.
Alas, the quest to completely replace electromechanical data storage
devices with solid state devices has been a holy grail for engineers from
the dawn of computing but unfortunately we are not about to achieve that
reality any time soon.
Keep in mind that engineering has one foot in science and the other
foot in economics. The cost in cents-per-bit for electromechanical
storage compared to that for non-volatile solid state storage is still
tilted in favour of electromechanical devices but the real fly in the
ointment is that typical demands for data storage are continually
rising.
So though a single SSD may have met all the storage requirements of a
modest sized company back in the 60's, even large arrays of them will
be insufficient for similar sized companies today.
Studies have shown that the amount of data being recorded is increasing
at 30 to 40 percent every year.
Advances in big-data analytics and AI-like processing of those data-sets
have resulted in companies amassing more information and keeping
it for longer than every before.
On the home front, fewer of us are buying DVD or BluRay disks
in favour of purchasing movies online and storing them on home servers.
All our photos are digital, we carry cameras in our pockets so we
take more photos and movies than ever before and we are more
likely not to print them but instead let them sit on a hard disk.
There have been many, many attempts to develop solid state
technologies that their creators hoped would one day eliminate
electromechanical storage.
As a researcher in the mid-80's, I had the opportunity to work with a
computing platform that had Intel bubble memory on it.
The bubble memory chips were very distinctive on the circuit board as
they were quite thick.
See
https://en.wikipedia.org/wiki/Bubble_memory
This was the technology that was being heralded as what would
replace magnetic hard drives.
Later that same year, I remember having farewell drinks at the pub
for a senior colleague who was departing to work on a new hush-hush
project in Silicon Valley. A few more beers later and he discretely
whispered that he was joining a startup to develop magnetoresistive
random-access memory. The promise was non-volatile data storage
but with semiconductor memory speeds.
The company did develop the technology, but it and other companies
came and went as new technologies were developed only to be eclipsed
by another.
These days of course we have FLASH and it has perhaps made the
biggest in-road into replacing hard drives particularly as the "C" drive
on a PC. A FLASH memory block, unfortunately, can only be erased
and re-written a finite number of times before it needs to be permanently
deallocated. Hence the requirement for "wear-levelling". A few lines of
code is all it takes to demonstrate the potentially short life of a FLASH
SSD if it were to be regarded as pure random access read-write storage.
So currently we tend to only store data on them that is unlikely to
change much.
One of the growing data storage challenges turns out to be DNA
sequencing.
With the cost of getting one's genome sequenced dropping, the increasing
utility DNA sequencing has for pathology and the requirements for
medical authorities to hold records for so-and-so number of years,
apparently the storage demands for the world's DNA sequencing
is becoming an ever increasing and urgent problem.
By 2025, some estimates are that up to one billion people will have
had their DNA sequenced.
Apparently compressing DNA data to some more desirable size is
challenging.
An article which appeared in the September 2018 issue of the Institute
of Electrical and Electronics Engineers (IEEE) warned that unless
researchers solve the looming DNA data compression problem,
biomedical science could stagnate.
See page 27 :-
https://spectrum.ieee.org/ns/Blast/S...m_2018_INT.pdf
Perhaps surprisingly to some, magnetic tape storage, in particular DAT,
has never gone away and continues to make strides in storage capacity.
A single robotic tape storage unit can hold up to 278 petabytes.
If the same data was stored on CD, it would take more than 397 million
of them and if they were stacked on top of each other they would
form a tower 476 km high.
Those numbers for off-the-shelf tape storage units gives one a sense
of the archival data storage requirements for a lot of organisations in the
world today.
A common mistake is to believe the types of storage requirements we
have on our home PC is somehow typical.
Perhaps one day semiconductor devices will completely replace
electromechanical storage. Perhaps quantum level devices.
But there is certainly no technology in existence today that shows any
hope of bringing an end to electromechanical storage.