Sign in with
Sign up | Sign in

HET MLC: Supercharged MLC Or SLC Lite?

Intel SSD 710 Tested: MLC NAND Flash Hits The Enterprise
By

SSD 320: MLCSSD 320: MLCHET MLCHET MLC

It's time to pick apart High Endurance Technology MLC, because the terminology is fairly new and, frankly, subject to some confusion. According to Intel, HET provides SLC-like write endurance. This is accomplished in two ways:

  1. Die-screening consumer MLC for marginally higher endurance.
  2. Increasing the page programming cycle (tProg).

Incidentally, these two traits are what define eMLC. In other words, HET is nothing but a marketing term. At the technology level, Micron tells us that die-screening (picking out the very best dies from a wafer) optimistically results in a two-fold write endurance boost. However, the company is citing an eMLC endurance spec that's six times higher than its consumer-grade MLC. Increasing the page programming cycle makes up the difference, as it potentially increases endurance by a factor of two or three.

3x nm Lithography
SLC
MLC
eMLC
Bits/Cell
1
2
2
Endurance (P/E Cycles)
100 000
5 000
10 000 - 30 000
ECC
8b/512B
24b/1KB
24b/1KB
tProg
0.5 ms
1.2 ms
2 - 2.5 ms
tErase
1.5 - 2 ms
3 ms
3 - 5 ms
Performance Over Time
Constant
Degrades
Degrades


Although die-screening seems like an easy (albeit cost-adding) way to cherry-pick the best pieces of memory without compromising performance, increasing the time it takes to program a page doesn't necessarily sound as sexy. The reason why relates back to the difference between MLC and SLC NAND.

Single-level cell flash stores one bit per cell. It is a single-bit binary system, either a "0" or a "1." MLC memory stores up to two bits per cell, so you're looking at four states to represent all possible combinations. Though that works out neatly on paper, there is a cost associated with increasing storage density.

Flash memory only has so much voltage tolerance. You can't just double the voltage to multiply the scale. Instead, you need more sensitivity between each state. This means more programming to manipulate a very precise amount of charge stored in the floating gate. MLC and SLC memory both operate similarly. However, MLC needs more precision in charge placement and charge sensing.

As P/E cycles are slowly consumed over the course of time, however, the read margins that determine the value of each cell start shrinking as a result of:

  • loss of charge due to flash cell oxide degradation
  • over-programming caused by erratic programming steps
  • programming of adjacent erased cells due to heavy read or writes


Consequently, over time, the drive experiences data retention problems and read-related errors. Basically, they wear out. That's not a problem on SLC-based drives because they only have one reference point. But MLC memory is completely different, which is why extending the page programming cycle has a pronounced impact on endurance.

In essence, additional time is spent sending a more precise charge to the memory cell. This increases the probability of writing to a cell within a smaller window, in turn creating much larger reference points and extending the amount of wear each cell can withstand. The result is higher endurance at the expense of less performance. And that, ladies and gentlemen, is the long explanation for the SSD 710's low 2700 IOPS random write rate.

Display all 30 comments.
This thread is closed for comments
  • -3 Hide
    whysobluepandabear , October 31, 2011 4:46 AM
    TLDR; Although expensive, the drives offer greater amounts of data transfer, reliability and expected life - however, they cost a f'ing arm and a leg (even for a corporation).

    Expect these to be the standard when they've dropped to 1/3rd their current price.
  • 4 Hide
    RazorBurn , October 31, 2011 6:48 AM
    To some companies or institutions.. The data this devices hold far outweighs the prices of this storage devices..
  • -3 Hide
    nekromobo , October 31, 2011 8:14 AM
    I think the writer missed the whole point on this article.

    What happens when you RAID5 or RAID1 the SSD's??
    I don't think any enterprise would trust a single SSD without RAID.
  • -2 Hide
    DjEaZy , October 31, 2011 8:29 AM
    ... glad, that i have vertex 3...
  • 0 Hide
    halcyon , October 31, 2011 10:04 AM
    Nice. Now let's see how many comments complain about the price. :sarcastic: 
  • 0 Hide
    halcyon , October 31, 2011 10:13 AM
    __-_-_-__with the reliability those have they will never ever find their way into any server

    My Vertex 3 has been very reliable and I'm quite satisfied with the performance. However, I've heard reports that some, just like with anything else, haven't been so lucky.
  • -2 Hide
    toms my babys daddy , October 31, 2011 11:50 AM
    I thought ssd drives were unreliable because they can die at any moment and lose your data, and now I see that they're used for servers as well? are they doing daily backups of their data or have I been lied to? ;(
  • 1 Hide
    halcyon , October 31, 2011 11:57 AM
    toms my babys daddyI thought ssd drives were unreliable because they can die at any moment and lose your data, and now I see that they're used for servers as well? are they doing daily backups of their data or have I been lied to? ;(

    SSDs are generally accepted to be more reliable than HDDs...at least that's what I've been lead to believe.
  • 0 Hide
    Onus , October 31, 2011 12:33 PM
    halcyonSSDs are generally accepted to be more reliable than HDDs...at least that's what I've been lead to believe.

    Yes, but when they die, that's it; you're done. You can at least send a mechanical HDD to Ontrack (or a competing data recovery service) with a GOOD chance of getting most or all of your data back; when a SSD bricks, what can be done?
  • 2 Hide
    CaedenV , October 31, 2011 12:48 PM
    nekromoboI think the writer missed the whole point on this article.What happens when you RAID5 or RAID1 the SSD's??I don't think any enterprise would trust a single SSD without RAID.

    The assumption is that ALL servers will have raid. The point of this article is how often will you have to replace the drives in your raid? All of that down time, and manpower has a price. If the old Intel SSDs were about as reliable as a traditional HDD, then that means that these new ones will last ~30x what a traidional drive does, while providing that glorious 0ms seek time, and high IO output.
    Less replacement, less down time, less $/GB, and a similar performance is a big win in my book.
    toms my babys daddyI thought ssd drives were unreliable because they can die at any moment and lose your data, and now I see that they're used for servers as well? are they doing daily backups of their data or have I been lied to? ;(

    SSDs (at least on the enterprise level) are roughly equivalent to their mechanical brothers in failure rate. True, when the drive is done then the data is gone, but real data centers all use RAID, and backups for redundancy. Some go so far as to have all data being mirrored at 2 locations in real time, which is an extreme measure, but worth it when your data is so important.
    Besides, when a data center has to do a physical recovery of a HDD then they have already failed. The down time it takes to physically recover is unacceptable in many data centers. Though at least it is still an option.
  • 5 Hide
    CaedenV , October 31, 2011 12:52 PM
    Oh! I almost forgot; GREAT review Andrew! I learned a lot on this one.
  • 3 Hide
    Anonymous , October 31, 2011 2:09 PM
    Lied to about what? And who are THEY? ... Life expectancy of SSD vs. Standard Harddrive? Thats always unknown, every unit is an animal unto itself. SSD's don't suffer mechanical issues however putting them ahead in my mind. Backups are determined by how much time you can afford to loose business-wise, how much data you have and how long it takes to recover to a point you backup at last. maybe your data is too valuable to have lost. In that case Mirror and even copy to a DR site, maybe even live. Best thing would probably be to trust your IT guy because you kinda seem lost :) 
  • 0 Hide
    halcyon , October 31, 2011 2:10 PM
    jtt283Yes, but when they die, that's it; you're done. You can at least send a mechanical HDD to Ontrack (or a competing data recovery service) with a GOOD chance of getting most or all of your data back; when a SSD bricks, what can be done?

    Its funny you mention that. Ontrack purports that they are quite adept at recovering SSDs.
  • -1 Hide
    mt2e , October 31, 2011 2:15 PM
    profit margins must be huge for a product with a simple memory swap
  • 0 Hide
    phate , October 31, 2011 7:04 PM
    So what's the difference between this and the P400e?
  • -2 Hide
    ruddenberg , October 31, 2011 7:40 PM
    Andrew Ku! Get the facts correct please !!!!

    Intel® SSD 710 Series 300/200/100GB
    Random Read (8GB Span) = no info
    Random Read (100% Span) = 38500/38500/38500 IOPS
    Random Write (8GB Span) = no info
    Random Write (100% Span) = 2000/2700/2300 IOPS

    Intel® SSD 320 Series 600/300/160/120/80GB
    Random Read (8GB Span) = 39500/39500/39000/38000/38000 IOPS
    Random Read (100% Span) = 39500/39500/39000/38000/38000 IOPS
    Random Write (8GB Span) = 23000/23000/21000/14000/10000 IOPS
    Random Write (100% Span) = 150/400/600/400/300 IOPS

  • -3 Hide
    cmartin011 , October 31, 2011 7:51 PM
    This is the best intel could come up with? i know reliability is important and all, but make the performance worth the price. at those dollar $ a much quicker PCI express solution could be afforded with some sorta redundant feature build in.
  • 1 Hide
    acku , October 31, 2011 9:16 PM
    ruddenbergAndrew Ku! Get the facts correct please !!!!Intel® SSD 710 Series 300/200/100GBRandom Read (8GB Span) = no infoRandom Read (100% Span) = 38500/38500/38500 IOPSRandom Write (8GB Span) = no infoRandom Write (100% Span) = 2000/2700/2300 IOPSIntel® SSD 320 Series 600/300/160/120/80GBRandom Read (8GB Span) = 39500/39500/39000/38000/38000 IOPSRandom Read (100% Span) = 39500/39500/39000/38000/38000 IOPSRandom Write (8GB Span) = 23000/23000/21000/14000/10000 IOPSRandom Write (100% Span) = 150/400/600/400/300 IOPS

    Read page 8. we covered that already.
  • 1 Hide
    campb292 , November 1, 2011 1:53 AM
    I find all the comments about data recovery very bizarre. What data would someone supposedly keep on a SSD (or HDD for that matter) that meets a threshold to warrant expensive data recovery in the event of failure, but not so sensitive to warrant a backup?

    My important info has a fresh original image and 2 daily backups that automatically create 12 hours apart. It takes about 5 minutes each and costs 29.99 a year. Come on people.
  • 0 Hide
    beenthere , November 1, 2011 1:55 AM
    This looks like more "experimenting" to see what enterprise will tolerate than a technical breakthrough. So far neither consumer grade nor the Intel 710 enterprise SSD impress me for performance, reliability and compatibility. It's certainly a painfully slow development on SSDs. Using consumers to beta test these drives is pretty unscrupulous IMO.
Display more comments