This week Intel announced the Optane SSD DC P4800X Series, new enterprise SSDs based on the company’s 3D XPoint memory technology which Intel says is the first new memory technology to be introduced since 1989. The technology was introduced to fill a price/performance gap that might impede Intel’s sales of high-performance CPUs.
Intel was all aglow with the promise of performance, claiming that the newly-released SSDs offer: “Consistently amazing response time under load.”
Since the early 1990s Intel has realized that it needs for the platform’s performance to keep pace with the ongoing performance increases of its new processors. A slow platform will limit the performance of any processor, and if customers don’t see any benefit from purchasing a more expensive processor, then Intel will be unable to keep its processor prices high.
Recently NAND flash SSDs have helped Intel to improve the platform’s speed, as did the earlier migration of Continue reading
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
On January 12 IBM announced some very serious upgrades to its DS8000 series of storage arrays. Until this announcement only the top-of-the-line model, the IBM System Storage DS8888, was all-flash while the less expensive DS8886 and DS8884 sported a hybrid flash + HDD approach. The new models of the DS8886 and DS8884 are now also all-flash.
But that’s not all: Every model in this product family has been upgraded.
The original DS8000 systems used a module called the High Performance Flash Enclosure (HPFE) for any flash they included, while these newer models are all based on HPFE Gen 2. While the original HPFE was limited to a maximum capacity of 24TB in a 1U space, the larger 4U HPFE Gen 2 can be configured with as much as 153.6 TB, for more than six times the storage of the previous generation. By making this change, and by optimizing the data path, the Gen 2 nearly doubles read IOPS to 500K and more than triples read bandwidth to 14GB/s. Write IOPS in the Gen 2 have been increased 50% to 300K, while write bandwidth has been increased by nearly 4x to 10.5GB/s .
This kind of performance opens new Continue reading
Micron has announced a new line of Enterprise SSDs that it has named the 5100 family. The three members of the family are designated by different suffixes: 5100 ECO, 5100 PRO, and 5100 MAX, as listed in the table below.
The three models support the same maximum read IOPS performance, but have a wide range of write IOPS figures, endurance (measured in DWPD = Drive Writes per Day), and maximum capacities.
All of these SSDs are based on Micron’s 3-bit 3D NAND. Micron has been aggressively ramping its 3D NAND technology since it began shipments in earnest last June.
The three SSD models are designed using the same fundamental firmware architecture, which Micron has named FlexPro, to yield consistent performance and reliability across the family, and with the hopes that customers will be able to qualify all three models in a single effort, which would provide one more reason for users to source their Continue reading
Yesterday IBM unveiled a sweeping update of its existing flash storage products. These updates cover a range of products, including IBM Storwize All Flash arrays: V7000F, V7000 Gen2+, and V5030F, the FlashSystem V9000, the IBM SAN Volume Controller (SVC), and IBM’s Spectrum Virtualize Software.
The company referred to this effort as a part of a: “Drumbeat of flash storage announcements.” IBM has a stated goal of providing its clients with: “The right flash for the right performance at the right price.”
IBM’s representatives explained that the updates were made possible by the fact that the prices of flash components have been dropping at a rapid pace while reliability is on the rise. The SSD Guy couldn’t agree more.
Here’s what IBM announced:
Starting from the low end and moving up, the V5030F entry-level/midrange array is an Continue reading
This is an update of the same survey we ran in 2012. We want to see how things have changed over the past four years.
Please click HERE and let us know what kind of storage performance you need. Even a hunch is good.
Let me back up a little – they’re not really slow. When Intel compared its standard NAND flash based PCIe SSD to a similar SSD based on 3D XPoint memory, the XPoint model ran 7-8 times faster, which is very impressive. Intel demonstrated that at the Intel Developer Forum (IDF) last August and several times since then.
But Intel and Micron have been boasting since its introduction that 3D XPoint Memory is 1,000 times as fast as NAND flash. How do you get from a 1,000 times speed advantage down to a speed improvement of only 7-8 times?
That’s what the graphic in this post will explain. The small rendition above is just Continue reading
Some time ago Objective Analysis ran nearly 300 standard benchmarks on a PC with varying amounts of flash and DRAM and found that a dollar’s worth of flash provided a greater performance boost than a dollar’s worth of DRAM once the DRAM size grew above a certain minimum (1-2GB) depending on the benchmark.
You might wonder how this could possibly be true. Everyone knows that best way to improve any computing system’s performance is to add DRAM main memory. How could flash, which is orders of magnitude slower than DRAM, provide a bigger performance boost than DRAM?
It all makes sense if you think of the DRAM of something that is there only to make the HDD look faster. More is better, but if you can use a little less DRAM and add a large flash memory layer then disk accesses appear to speed up even more.
The benchmark data and the price/performance findings that are 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