I’m in Silicon Valley at the Flash Memory Summit. The big news so far: Intel and Micron announced they are sampling an 8 GB, 3 bit-per-cell (3bpc) NAND flash.
Commercial availability is expected in Q4, but don’t be surprised if that date slips. Process tweaks needed to reach full spec parts in the next 6 months aren’t guaranteed.
Bang/$
3bpc means 50% more capacity per $. That’s good.
The tradeoff: it will only handle about 1,000 writes before failing. Which, as I pointed out 4 years ago in an early post about server flash SSDs, isn’t much of a problem when you have lots of capacity.
This flash will 1st go into products like USB thumb drives or SD cards, where you won’t write to it 1,000 times anyway. With wear-leveling you’ll have to write 8,000,000 MB of data to an 8 GB part before it croaks. That is 2 million 4 MB JPEGs. That’s a lot of snapshots.
Since most of these will go into devices that hold 16, 32 or even 64 GB of flash, multiply 2 million by 2, 4 or 8 and it is obvious that casual users will never wear out 3bpc flash.
Write performance will suffer as well, but how much remains to be seen. The effect will depend on how many flash chips are on the device: flash controllers write data in parallel, so the more chips the faster the write.
The StorageMojo take
3bpc products will have a ripple effect on other flash parts. Price sensitive products – most flash parts – will want to move to 3bpc ASAP, but the volumes won’t be there. But as production ramps, 2bpc flash will face big price pressure as vendors with older fabs try to keep them running.
Translation: expect that flash prices will resume their downward path after 3 quarters of flat prices. Yay!
Comments welcome, of course. SSD vendors shipped about 10 million SSDs in the last 12 months. Sounds good, but tiny compared to over 500 million hard drives.
With SSD some applications and OS need to be rethink to avoid writes.
I just bought a new HP desktop and notice my harddisk activities was high event during idle. I used a freeware tool that monitor disk activities and found out it was HP Advisory utility who constantly log every 5 seconds. This is not the only software who cause high activity for no reason. Microsoft has several of them along with anti-virus.
The good thing now is CPU and other hardware vendors acquire anti-virus to be place right in the datapath rather than the virus land on my harddrive or memory.
Still too many bad activities by applications that waste my harddisk life…and my time.
Frankly, I can’t think of many situations where I’d like to trade 1.5 * extra capacity for 0.1 * durability, apart from sneakernet thumb drives, which spend most of their time sitting around unplugged. You can write more passes to an LTO-5 or T10K tape than you can to devices made from 3-bit flash!
Like Visiotech, I have seen enough scares w.r.t. limited lifetime components getting automatically aged to failure – laptop hard drives getting their heads loaded/unloaded every few seconds when running perfectly normal Linux distros is the example that comes to mind – that I would rather not assume perfectly well-behaved behaviour from all the components in my system, including the wear levelling firmware on the flash device itself.
I’d really like to see hybrid SLC/MLC devices which absorb most of the small and “scrubby” writes onto durable SLC, with migration out to MLC as the data ages, or direct to MLC for big writes. Then you could have your filesystem journal and data going to the same SSD with good performance and durability for both cases.
RE: “I’d really like to see hybrid SLC/MLC devices”
So would I.
Excellent paragraph. Right to the point.
When I talk to designers and Storage people they seem in lock-step with the “a write is a write is a write” approach. This means they want one super-quick write and not have to worry about accesses anymore. Speed and Capacity seem to be the defining parameters. Those parameters were about the limit for “Rotating Rust”.
SSD/Flash provides a whole new world of possibilities.
What if we had a BackBlaze box, http://storagemojo.com/2009/09/01/cloud-storage-for-100-a-terabyte/, that could be all-of-a-kind or mixed “Rotating Rust”, SSD, pluggable Flash (easily plug-n-play for archiving Information) and was User configurable? Needs to be very modular. Quite a design challenge. Texas Microsystems also has the technical ability to produce one of these.
I’m not sure who has this type of “SSD of the Future” on the drawing board.
What this type of box gives is a feature/function set that makes tiering, layering, replication, and a host of other Information Management Strategies with tools very available and configurable.
Being configurable extends the useful life of the boxen. Some people want planned obsolesence or “throw-away” Storage.
Maybe we want Storage that is Mobile, Agile and Transparent?
Push-pull-click-click we just redeployed several Petabytes of Storage (Mobile) and re-configured it on the fly (Agile) to support a new Line of Business showing explosive growth and we did it all from a browser on a desktop/laptop (Transparent). It was even preloaded with test suite Information and run through the Test Suite for that Line of Business before being brought on line. A layer of SSD/Flash allows this to be done.
This brings up the point of where do we want the intelligence?
In the boxen (fastest) or external (latency?). We might need a new “bus” structure defined to handle this.
I hope that there is little or no slower read/write with 3BPC vs. 2BPC Looks like I have a reason to wait for the next big SSD tech.