SSDs use a huge number of internal parameters to achieve a tricky balance between performance, wear, and cost. The SSD Guy likes to compare this to a recording studio console like the one in this post’s graphic to emphasize just how tricky it is for SSD designers to find the right balance. Imagine trying to manage all of those knobs! (The picture is JacoTen’s Wikipedia photo of a Focusrite console.)
Vendors who produce differentiated SSDs pride themselves in their ability to fine-tune these parameters to achieve better performance or endurance than competing products.
About a year ago I suggested to the folks at NVMdurance that they might consider applying their machine learning algorithm to this problem. (The original NVMdurance product line was described in a Memory Guy post a while ago.) After all, the company makes a machine learning engine that tunes the numerous internal parameters of a NAND flash chip to extend the chip’s life while maintaining the specified performance. SSD management would be a natural use of machine learning since both SSDs and NAND flash chips currently use difficult and time-consuming manual processes to find the best mix of parameters to drive the design.
Little did I know that NVMdurance’s researchers Continue reading
Sometimes it’s enlightening to compare several viewpoints on similar data. At yesterday’s SNIA Persistent Memory Summit a number of presentations provided interesting overlapping views on certain subjects.
One of particular interest to The SSD Guy was latency vs. IOPS. Tom Coughlin of Coughlin Associates and I presented the findings from our recently-published IOPS survey report and in Slide 19 displayed the basic chart behind this post’s graphic (click to enlarge, or, better yet, right-click to open in a new tab). This chart compares how many IOPS our respondents said they need for the storage in their most important application, and compared that to the latency they required from this storage. For comparison’s sake we added a reference column on the left to roughly illustrate the latency of various standard forms of storage and memory.
You can see that we received a great variety of inputs spanning a very wide range of IOPS and latency needs, and that these didn’t all line up neatly as we would have anticipated. One failing of this chart format is that it doesn’t account for multiple replies for the same IOPS/latency combination: If we had been able to include that the chart would have shown a clearer trendline running from the top left to the lower right. Instead we have a band that broadly follows that trend of upper-left to lower-right.
Two other speakers presented the IOPS and latency that could be Continue reading
In essence it’s because both the HDD industry and the semiconductor industry have set goals for themselves to achieve 30% average annual price reductions. If they are both on the same trajectory, and if there’s an order of magnitude difference between HDD and SSD prices today, then there will be an order of magnitude difference in the future as well.
The 30% average annual decline in SSD prices has a convenient name: Moore’s Law. Although there’s no physical, economic, or other restriction behind Moore’s Law (so it’s not really a law at all) it serves as a guide for the industry. Chip makers set their sights at doubling the number of transistors on a chip every couple of years, and this equates to average annual price decreases of 30%.
The HDD business also Continue reading
This Sunday (Sept. 20, 2015) I will be presenting my company’s findings on the 3D XPoint memory that was introduced by Intel and Micron in July. I will be speaking at the Storage Networking Industry Association (SNIA) Storage Developer Conference (SDC) Pre-Conference Primer. You can click the name to be taken to the agenda.
This won’t be the only talk about persistent memory technology at the conference. Prior to my presentation storage consultants Tom Coughlin and Ed Grochowski will give an overview of advances in nonvolatile memories, and following my presentation will be two Intel talks.
Intel will be covering this new technology a lot during the conference. Of a total of 120 presentations at the conference and pre-conference primer, Intel will be presenting nine, seven of which directly name persistent memory or nonvolatile memory in the title. Other firms will also be talking about NVM: AgigA, Calypso, HP, Pure Storage, and SMART Modular. Even Microsoft alludes to it in a couple of its presentation titles. Persistent memory is a hot issue.
So, the question for readers of The SSD Guy blog is: “Will this do away with SSDs?”
This is a question that was Continue reading
From time to time IT managers ask The SSD Guy if there’s an easy way to compare SSDs made with MLC flash against those made using eMLC flash. Most folks understand that eMLC flash is a less costly alternative to SLC flash, both of which provide longer wear than standard MLC flash, but not everyone realizes that eMLC’s superior endurance comes at the cost of slower write speed. By writing to the flash more gently the technology can be made to last considerably longer.
So how do you compare the two? OCZ introduced MLC and eMLC versions of the same SSD this week, and this provides a beautiful opportunity to explore the difference.
As you would expect, the read parameters are all identical. This stands to reason, since Continue reading
I have to admit that it’s embarrassing when The SSD Guy misses something important in the world of flash storage, but I only recently learned of a paper that Baidu, China’s leading search engine, presented at the ASPLOS conference a year ago. The paper details how Baidu changed the way they use flash to gain significant benefits over their original SSD-based systems.
After having deployed 300,000 standard SSDs over the preceding seven years, Baidu engineers looked for ways to achieve higher performance and more efficient use of the flash they were buying. Their approach was to strip the SSD of all functions that could be better performed by the host server, and to reconfigure the application software and operating system to make the best of flash’s idiosyncrasies.
You can only do this if you have control of both the system hardware and software.
The result was SDF, or “Software-Defined Flash”, a card that Continue reading
Erik Logan of Pogo Linux sent me a link to an amusing & informative video he and his company put together called A Brief History of SSDs. In the video Erik (pictured) tells of Pogo Linux’ experience with SSDs.
The company has a lot of hands-on SSD experience: Pogo Linux ships servers and storage and has ramped SSD shipments (as a percent of all drives) from single digits three years ago to the point where SSDs now account for 31% of all drives they ship. Erik shared with me that: “Sorting through the Continue reading
Tom Coughlin and I have just released a new report that helps shed a lot of light on a pretty challenging subject: We asked nearly 200 IT managers to tell us how much storage performance their systems require. They provided candid replies about their IOPS, latency, and capacity needs for a number of leading applications.
SSD spec sheets, might lead you to believe that power is just not an issue. For example, Samsung lists the power use for a 512GB 830 SSD at 0.127W (typical) for “Active Power Use”. This implies very low demands on the system power supply.
If you do some more research, you find that the peak power usage is a lot higher. AnandTech, in a review article reports sequential write power draw at 5.14W and random write power draw at 5.8W. In that 2.5” SSDs use the 5V power rail exclusively, this is more than Continue reading
In his Flash Memory Summit keynote on Wednesday, Micron VP and Chief Memory Systems Architect Ed Doller made a compelling demonstration of the power and performance advantages of a new approach to computing.
With true showmanship, Doller had his co-workers hand out buttons with LED lights to the entire audience. The LEDs in these buttons were either green or blue, with the colors randomly dispersed among the crowd. Doller asked the entire audience to turn on their lights, then called one row of the audience to file up to the stage so he could determine whether each person’s button was blue or green.
He pointed out that this was like having a single CPU check the contents of a drive. He then asked why things should work this way – wouldn’t it be more sensible to Continue reading