backtop

Print E-mail del.icio.us 30 comment(s) - last by Shadowmaster62.. on Oct 10 at 3:34 PM
Seagate and JEDEC working together for SSD industry standard

Seagate is the largest hard drive maker on the planet. Hard drives use platters of magnetic storage that rotate at different speeds for data storage. Variations of this technology have been used for years for data storage in both consumer and enterprise environments.

The technology behind HDDs is tried, true, and comfortable for consumers and enterprise customers alike. New technology in the form of the solid state drive (SSD) has slowly trickled into the environment with the promise of much higher performance than the traditional platter-based HDD can provide.

The trade off with any new technology is that durability can be an unknown, which leads enterprise customers looking for storage for crucial systems to be weary of new technology. Seagate previously stated that in 2009 it intends to enter into the SSD market. Before jumping into the market with both feet, Seagate is working on alleviating fears in what it believes to be the key SSD market -- high demand enterprise markets.

To dispel fears Seagate believes that large enterprise customers need to be convinced of the durability of the SSD. To convince these customers Seagate is working to develop an industry standard for durability with the JEDEC standards body. Seagate's Rich Vignes is quoted by CNET News saying, "As companies like Seagate start to demonstrate field-proven reliability and endurance in enterprise applications, we'll overcome those (solid-state drive) endurance fears."

Seagate also says that it has decades of experience with error correction in storage media and that it can apply its experience with error correction from HDDs to new SSD products. Once the fears of the enterprise user can be quelled, Seagate sees speed driving adoption of the SSD. Industry Analyst Gregory Wong said, "SSDs have 100 times better random IOPS than HDDs."

Seagate is confident in its ability to enter into the SSD market and gain traction as a preferred supplier. Vignes said, "While for some companies, it's a new market and a new product, for us, it's an existing market, new product."

Comments     Threshold


This article is over a month old, voting and posting comments is disabled
Fake standard?
By the goat on 10/9/2008 11:58:19 AM , Rating: 4
This endurance rating will be just as useful as the MTBF number on current hard drives. The companies will manipulate data to show whatever rating they want to advertise.

The warranty is what matters not some artificial endurance rating. Just like hard drives, solid state drives will be replaced while under warranty. After the warranty expires you are SOL regardless of the MTBF rating.




RE: Fake standard?
By Samus on 10/9/2008 12:16:54 PM , Rating: 3
I think this has more to do with "detect before failure" not mean time between failure. Most SSD's don't have any system in place to detect block decay, resulting in unpredictable data loss.

Basically Seagate is inventing SMART for the SSD. Although MTBF is usually manipulated as you said, there is no arguing that SMART is a useful feature as it has often predicted drive failure when used properly (IBM/Dell workstations use it well)


RE: Fake standard?
By bhieb on 10/9/2008 12:33:13 PM , Rating: 4
Please correct me if I'm wrong.

My understanding is that SSD's are actually safer. Once written data is relatively safe, the problem lies in writes. Intel's approach is that if a write fails that block is marked as bad and it is written somewhere else. Therefore no data will theroretically ever be lost. And utilities can warn you when your running out of "good" blocks. It makes Raid 0 seem much more appealing since the likelyhood of a bad block is remote at best.


RE: Fake standard?
By Doormat on 10/9/2008 12:44:43 PM , Rating: 2
From what I read about Intel's SSDs, you're right.

Read/write cycles aren't the issue, is dead chips are - solder joints breaking down (tin whiskers) or something inside the chip going bad.


RE: Fake standard?
By therealnickdanger on 10/9/2008 12:48:34 PM , Rating: 2
That seems right - or at least in line with what I've read elsewhere.


RE: Fake standard?
By Oregonian2 on 10/9/2008 2:23:19 PM , Rating: 3
Can Intel's detect write failures without having to do a read-back cycle, or is that what they're doing? IOW does their method of detecting a write-failure guarantee that reading data back will be correct, or does the algorithm just guarantee that write-data with a write-failure will indeed read back incorrectly (very very different)?


RE: Fake standard?
By mathew7 on 10/10/2008 6:59:16 AM , Rating: 2
Actually you're wrong. While the old data is not directly overwritten, it is written to a new sector which is mapped to that logical sector. The OS only sees the logical sectors, so while you may still have the old sector on the drive, there is not (standard/easy) way to access it. It's similar to the spare sectors on HDDs, except that the SSD have much more of them. I would not be surprised to be 10-20% more sectors than the OS sees. These are specifically reserved for wear leveling.
Also, RAID 1 does not protect you only from bad sectors, but also from total drive failure (like head crash, motor failure). This is also applicable to SSDs.


RE: Fake standard?
By the goat on 10/9/2008 12:58:15 PM , Rating: 2
quote:
I think this has more to do with "detect before failure" not mean time between failure.


No you are wrong. If you were to read more about this you will see Seagate wants jedec to come up with a way to express the "endurance" of an SSD. This will be a metric used for advertising. This is not something designed into the drive to detect failures.

The is nothing stopping SSD drives from using the exact same SMART monitoring system that HDDs use. They both plug into an SATA bus.


RE: Fake standard?
By Samus on 10/10/2008 9:45:22 AM , Rating: 2
There are currently no SMART parameters that would function with SSD architecture other than operating temperature, power cycles and other generic, useless information.

I agree Seagate wants JEDEC to help them with a MTBF 'rating' for SSD's, but if YOU read the CNET article is does state Seagate is working on a reliability 'detection' system.


RE: Fake standard?
By FATCamaro on 10/9/2008 1:09:18 PM , Rating: 2
These are intended for marketing purposes. To make people feel secure.


RE: Fake standard?
By Oregonian2 on 10/9/2008 2:26:58 PM , Rating: 2
Yes, the article explicitly says this. Big industrial customers want to be assured long term performance before putting their information-butt on the line. If there are data disasters from SSDs, they can point to the numbers and pass the blame on -- else they get to receive the bolt of lightning themselves.


RE: Fake standard?
By winterspan on 10/9/2008 5:43:44 PM , Rating: 2
quote:
Most SSD's don't have any system in place to detect block decay, resulting in unpredictable data loss.


That doesn't sound correct. From all the research I've read, when the blocks wear out they just become read-only -- you don't lose data. So in that respect, they are actually much safer than a conventional harddrive.

Also, although SSD specifications and quality vary *WIDELY* depending on manufacturer, even the lower-cost MLC variants (with 1/10 the life of SLC) will easily outlast the life of a laptop for an average user. And remember, reading doesn't have a wear effect on the cells, only writing does. So unless you are moving 100's of gigabytes around on the drive each day, then an MLC drive should be fine. for high-throughput server use, obviously SLC will be more durable.


RE: Fake standard?
By mathew7 on 10/10/2008 7:22:28 AM , Rating: 2
It becomes read-only, but there is no usable data (because it's damaged) and it's hidden to the OS by the flash controller. Also it's an issue of when it's detected as bad (like too late).
The mostly used wear-leveling algorithm will free a sector when it's replaced (when that logical sector is written but it's written to a free sector). So even if you use 1% of the drive and never write to the rest, you can still wear-out that 1% of the drive. This is because the flash controller does not know which sectors are free according to the OS. So I think a monthly free space write scan like in scandisk could actually improve the wear leveling.


RE: Fake standard?
By omnicronx on 10/10/2008 2:52:20 PM , Rating: 2
Yes because everyone buys retail... Warranty means little to nothing to those that buy OEM products, thats where the MTBF number does come into play.


Intel already set the standard.
By therealnickdanger on 10/9/2008 12:14:36 PM , Rating: 2
It's called the X-25M. Put up or shut up, Seagate.




By FrankoBeans on 10/9/2008 12:57:24 PM , Rating: 2
Industry standards groups coordinate efforts and minimize wasted effort (where there is no potential for profit by being the sole source), i.e. USB standards. These discussions improve compatibility and keep technology a little less complicated.


RE: Intel already set the standard.
By geddarkstorm on 10/9/2008 1:10:44 PM , Rating: 3
Funny, considering how Seagate was bashing SSDs not too long ago.


RE: Intel already set the standard.
By Reclaimer77 on 10/9/2008 1:19:47 PM , Rating: 2
Lol exactly.

Hey maybe if we get boast about our self written " INDUSTRY STANDARDS " nobody will notice that we stupidly shunned this new technology and won't buy all the already superior SSD's already on the market while we struggle to release our INDUSTRY STANDARD SSD's in two years.


RE: Intel already set the standard.
By Oregonian2 on 10/9/2008 2:37:55 PM , Rating: 2
The SSD's won't be "industry standard", just some of the specs.

What if a company advertises that their SSDs last a minimum of 100 years? Unknown is that the company's internal definition of longevity is that if they build ten thousand units, the number signifies the length of time it takes for the very last one to fail -- and only when they are used for no more than one week per year. Another company advertises the same thing but their internal definition is how long until the first one fails and they run it at max temperature and running continuously. Same number advertised, they must be the same!

Standard specs are a good thing even if it isn't thought highly of by critics.


RE: Intel already set the standard.
By Reclaimer77 on 10/9/2008 3:11:09 PM , Rating: 2
And you don't see a slight issue with company setting the standard also being the ones manufacturing and profiting from the SSD's ??

quote:
Standard specs are a good thing even if it isn't thought highly of by critics.


Sure. If its done by already established and esteemed INDEPENDENT standards organizations.


By Oregonian2 on 10/9/2008 4:13:06 PM , Rating: 2
Won't happen here, nor has it ever happened AFAIK for all computer/electronic standards. Even in those organizations that you think "independent", the people doing the work are all representatives for companies who have an interest in the standard and are paying their salaries.