SSHD with upto 4Tb available ,may change trend

[bvreddy]

Its from seagate , Hybrid hard disk ( ssd + hdd = SSHD ).

Anybody knows how its combined two technologies......

B.V.Reddy

[HaQue]

there is this cool website that can tell you all kinds of things, you should check it out some time:

https://www.google.com

or look here http://www.seagate.com/au/en/internal-hard-drives/solid-state-hybrid/desktop-solid-state-hybrid-drive/

[MarkGreenwood (KE)]

[data-medics]

there is this cool website that can tell you all kinds of things, you should check it out some time:

https://www.google.com

Wow, you can actually cut the sarcasm with a knife it's so thick. Haha!

To give a simple answer to the OP's question. The SSD portion basically just acts as a large cache for the most commonly accessed sectors. At first it's just blank, but after a sector block is accessed a certain number of times it's moved to the SSD. Then later on, as different areas are used more and more, data is moved back and forth so the most frequently accessed is on the SSD. It's all handled internally by the SSHD.

[data-medics]

I think in most cases the data is all still stored on the platters as well, and just duplicated into the SSD for faster access, but I can't verify this.

Also a portion of the SSD acts as a write cache, so you can write data to it faster, and it writes it to the platter later when it's less busy.

[bvreddy]

Thanks data-medics

From recovery point of view.....If SSD memory chips are maintaining duplicate copy of most frequently accessed data then we can retrieve entire data from plotters..need not apply SSD recovery process..

In case of a plotter issue then recovery from SSD chips may not be possible i guess, because cache maintains data in its own way but not sequentially.

[HaQue]

I think there are a few different schemes in operation, probably every combination of HDD/SSD would exist somewhere.. DATA+Cache, DATA+DATA, DATA=Duplicate oft used DATA etc.

I am wondering when big-iron storage tech will filter into consumer products. Stuff like De-Duplication tech where common data exists once, and any subsequent copy is replaced by a pointer to that data.

I envisage it working something like this where you have some data, then the copies are just pointers to the start and the length of bytes.:

Things like file headers, database values like first names.. could contain duplicates
needless to say if the first copy gets lost, from a dead nand or head crash, you are going to be S.O.L.

This sort of thing is figured out by very complex algorithms, and the days of looking at a disk with a full, if chopped up, copy of your data in unencrypted format I think are numbered.

But if you save that much space, 128bytyes as opposed to 228bytes, well over 50% saving in this simple example, then it is going to come before DR, it should also be a lot faster as much less data to hold in a cache.