[Discuss] SSD: enterprise vs. consumer Flash

[tkoker at gmail.com (Tony Koker)]

Hi,

   I do storage as my specialty (last 7 or 8 years) and at my primary
source of income, currently, we have several different SSDs in use
(Production or POC).

   A big player like EMC has factored SSD to mostly be plug-in replacements
for the other disk drives by physical form factor, protocol and interface,
and, mostly use MLC (Multi Level Cell) technology, in various degrees, and
complex cache and controllers for balancing trade-offs in capacity versus
access time (high speed and low latency). They know they can charge a
premium for this 30x times faster than disk (for reads, anyway), and the
extra technology they have to include to make them plug-in replacements,
especially given the unknowns concerning real-world reliability, given
their short period of existence. They don't wear out mechanically, like a
spinning disk drive might, but the cells die (won't flash), so they lose
the ability to be written-to, so their life-time is dependent upon the
quantity of writes that are performed.

   Violin Memory, Inc describes two of their offerings thusly:

*Performance Flash Technology*
The Violin 6616 is based on Single Level Cell (SLC) flash memory and
optimized for high IOPS and low latency while still providing robust RAID
protection, ultra-low response times, high transaction rates, and real-time
queries of large datasets.

*Capacity Flash Technology*
The Violin 6232 is based on Multi Level Cell (MLC) flash memory and
optimized for high capacity while still providing robust RAID protection.
Violin?s unique vRAID-enabled MLC results in the fastest MLC Memory in the
industry.
NAND and NOR difference is about flashing, or erasing; NAND has a page
often much bigger, while NOR's page may be as small as a byte. But the
issue of NAND vs NOR also focuses on density vs flashing and/or writing
(erasing or read, erase, re-write cycle of controller):

"The high density NAND type must also be programmed and read in ... blocks,
or pages, while the NOR type allows a single machine word (byte) to be
written or read independently." Wikipedia ("Flash Memory")."

Historically NAND is mostly in use for SD, USB (memory cards), flash and
SSDs (where blocks, page, sector, track, cyclinder, etc apply easily),
while NOR is more likely to be found where you'd previously find
EPROMs, EEPROMs
or battery-powered static RAM use cases like embedded systems code and data.

Tony Koker


On Sun, Jun 3, 2012 at 1:44 AM, Tom Metro <tmetro+blu at gmail.com> wrote:

> Richard Pieri wrote:
> > Tom Metro wrote:
> >> Edward Ned Harvey wrote:
> >>> The flash they use in USB sticks and SD cards is the same flash
> >>> they use in enterrpise hard drives.
> >> Are you sure about that?
> >
> > He's right. USB flash and SSD flash both use NAND flash as opposed to
> > NOR flash.
>
> Just because both use parts from the same technology family doesn't mean
> they are the same part.
>
>
> > ...if you compare Sandisk with Sandisk you'll find the same flash
> > chips inside.
>
> So you're saying if I crack open a Sandisk USB drive and a Sandisk
> enterprise SSD I'll find chips with identical part numbers inside? Call
> me skeptical...
>
>
> >> What makes an enterprise SSD controller more reliable? Or are you
> >> saying the controller itself isn't more reliable, but it implements
> >> hardware wear-leveling algorithms that makes the overall SSD more
> >> reliable?
> >
> > What makes a 3ware RAID card more reliable than a Marvel RAID card?
>
> I'm not so sure it is. Is the MTBF of the *card* (significantly) any
> different? I doubt it.
>
> I suspect he is talking about the reliability of the storage system
> improving as a result of a more sophisticated controller, but this
> wasn't clear, and thus I asked. I wouldn't have phrased it as the
> *controller* being more reliable.
>
>
> >>> Flash itself is dirt cheap.  What you're really buying is the
> >>> on-device flash controller, that maps the flash blocks to virtual
> >>> HDD blocks, and implements the USB/SATA operations.
> >> Are you sure about that?
> >
> > He's correct on this as well.  NAND flash is cheap.  Controller
> > circuits aren't so cheap, particularly when a small number of
> > manufacturers are investing heavily in performance wars.
>
> The Flash chips have commodity volume production in their favor (which
> can be significant). If the SSD's with high-end controllers are also
> using high-end Flash chips that are produced in smaller volumes, then
> this advantage is nullified.
>
> Beyond that, the economics of semiconductor fabrication would suggest
> this is incorrect, but I'll happily read reference material saying the
> contrary, if you have it.
>
>  -Tom
>
> --
> Tom Metro
> Venture Logic, Newton, MA, USA
> "Enterprise solutions through open source."
> Professional Profile: http://tmetro.venturelogic.com/
> _______________________________________________
> Discuss mailing list
> Discuss at blu.org
> http://lists.blu.org/mailman/listinfo/discuss
>