Solid-State, no moving parts Hard-Drives

Yes, I certainly can see this replacing spinning platters. Particularly for portable devices. I was surprised by your 4 million cycle MTBF, and I was right. Intel claims 16 million hours, not cycles. That seems a little high to me, particularly since that's just under 137 years of continuous use. Even if that's off by a factor of ten, I'd still be impressed. Now if we could only convince some code writers not to waste storage space....

That right. 4 million hours that was.

Hi Yuval,

Yes, absolutely. As this technology advances, and more people indulge in it, prices will drop and it will become desirable to more people. Because these SS (that's Solid State, not Stainless Steel for all the DWA (Down With Acronyms) people out there) have no moving parts, they are more reliable.

I think that doing away with the motor, bearings, R/W heads and magnets that traditional hard drives have would make the SS drives even cheaper.

I don't know anything about differences in access times, throughput, etc., but I believe that if more people would invest in this technology, these would be cheaper than the equivalent capacity rotary drives within a couple of years.

Mike

Redfred: 137 years for solid state devices is not considered impossible - most solid state devices (PCBs, discrete components, ICs, VLSI) theoretically have an unlimited time of operation provided their optimal temperature power and signal input. Not that it remains relevant for 137 years, when a year to five later are discarded as obsolete.

Mikerho: Thei lack of moving parts may explain the above, as well make a great shock-proof gear for military, sports, and critical use. Some SMT boards these days can withstand tens of G's when dropped or bumped. You can store your entire media library in the souls of your jogging-shoes, or hide a mediaplayer in your baseball bat...

Not very practicle examples, I know, just to illustrate...

There are two other issues that come into play with regard to solid state memory. One is data retention. Since the data is stored as electrical charges and the cells are not perfect insulators there is a mean time before the charge dissipates an looses the stored state. Another issue has to do with the manner in which the cells are erased. A relatively high voltage is applied to reset all the charges in order to "erase" the cells before they are re-written. Due to the very thin nature of the insulators that are used, erasure constitutes a stressor on them and they may eventually fail as a result.

Yes, these devices have a limited number of erase-write cycles, so expect this type of drive to fail sooner with an active paging partition or "swap" file. In addition, erase takes longer than write, and write takes longer that read. However, they are ideal as a (mostly) read-only operating system partition.

I think that they also incorporate some virtual addressing algorithms to spread frequent usage over the entire array to lessen the likelihood that you "wear out" one spot too quickly.

So I guess de-frag of one of these is not just pointless, but harmful.

I suppose so. Since they are random access devices you would not incur the track seek latency of a mechanical device. Probably the OS has to check for the next sector location on each sector anyway so there would really be nothing to gain except in the event that there were the opportunity to do continuous sequential reads. Otherwise you are just using up some of your r/w cycles to move things around for little or no gain in performance.

Yes, they mark invalid blocks that have not yet been erased, and write the new file versions in "virgin" blocks. When the disk is nearly full of invalid blocks, they are finally reused. This way erasing is not concentrated in any one place.

That makes sense. It would also improve write latency since they could actually block erase during inactive periods so that it didn't normally delay writes.

I had an SD card die on me and lost all the contents. So I don't get too excited about the infallibility of solid state memory devices. In 24 years I've never had a hard disc die. I did have to throw one away because I could not figure out how to get rid of the currupted remnants of Win 2000 pro on it; the result of my hitting the wrong combination of keystrokes and mouse clicks.

Ed Weldon

With most vendors selling silicon disks, a lifetime warranty is attached. They must be very confident in the reliability of that technology.

Again, as stated above, it doesn't prevent your device from becoming obsolete in a very few years...

SD ≠ SSD ??

SD is for Silicon Disk, and SSD is for Solid-State Disk - which are all the same

I'm not sure about the names and acronyms used here. If it has no moving parts, then I can't call it a Hard Drive. All of the solid state memory units that I've seen were rectangular, so I can't call it a Silicon Disk.

The MacBook Air™ has been shipping for quite some time (over a year) with nothing but solid state memory storage, and I'm aware of at least one regular contributor to CR4 who has 256GB of it in a MacBook Pro™. (not me, yet)...

It is going to be very interesting to see how long these things will keep their data intact.

I recently bought a 1TB usb drive... spent a couple of days copying some files to it.. then accidentally knocked it off the desk.. now it makes funny clicking sounds and won't read... so I think solidstate is excellent.. especially when computers are here that are on a usb 'memory stick' size device. (monitor not included)

Chris

It sounds like the solution is to skip the USB and go straight to Compact Flash, micro-SD or similar; solid state with a more rugged interface.

USB came about year2k because everyone was laughing at the term "Plug-And-Play" - mocking it to near-death.

The initial idea was to integrate an I/O interface for portable gadgets hooking for a momentary exchange of information such as cameras, cellphones, laptops etc.

The solution went gaga with something that never proved reliable. It was dodgy from the start, and it's as dodgy today as it ever was.

That solution had it's own standardised power-supply and a protocol to be installed, to enable an external device to be plugged and un-plugged instantly.

The problem was that the designed allowed for the data structure to be zapped by the power-supply, with horride consequences. This had to be eliminated for both plugging into a host, and unplugging. That task to eliminate the data being zapped, they decided to relay to the host - let the host save the device from doom - every conection.

In plugging to the host, their solution was to block the zap by the logic of the host, while the data-structure was established on the host (PC). IN unplugging, theit solution was to block the zap, again, by the host, until the device was no longer detected, by the host.

Only it caught on. It became a fad and went into territories never intendede like a fixed interface for a mouse, keyboards, printers - you name it.

Parking a permanent device on the UB port slows the whole PC because the port has to re-iterate every second for the presence of lack of a device - this is it's autodetect loop. This occupied endles stream of PU cycles and take it's tall on the whole system.

- sorry for my poor spelling - the spell-checker was out of commission

Hello
I done a compact flash memory based project for data acquisition and it was a disaster. At the moment what is in the market does not work very well for applications 24H.

Many sectors of the cards were damaged and doing a fsck gave different errors each time. We had to replace almost every year of operation. Until we see computers in stores without a hard drive (which takes a minimum of one year sold) will not be a good idea for developments, i think.

Be happy everybody

Escriviu text o una adreça d'un lloc web o bé traduïu un document.Cancel·laEscoltaLlegeix fonèticament Traducció castellà > anglès