Archive for the ‘flash memory’ Category
May the SandForce be with you
Nice writeup from Anandtech regarding the press release from LSI about it’s new 3rd generation flash memory controllers. The 3000 series takes over from the 2200 and 1200 series that preceded it as the era of SSDs was just beginning to dawn (remember those heady days of 32GB SSD drives?). Like the Frontier days of old, things are starting to consolidate and find an equilibrium of price vs. performance. Commidity pricing rules the day, but SSDs much less PCIe Flash interfaces are just creeping into the high end of the market of Apple laptops and soon Apple desktops (apologies to the iMac which has already adopted the PCIe interface for its flash drives, but the Mac Pro is still waiting in the wings).
Things continue to improve in terms of future-proofing the interfaces. From SATA to PCIe there was little done to force a migration to one or the other interface as each market had its own peculiarities. SSDs were for the price conscious consumer level market, and PCIe was pretty much only for the enterprise. You had pick and choose your controller very wisely in order to maximize the return on a new device design. LSI did some heavy lifting according to Anandtech by refactoring, redesigning the whole controller thus allowing a manufacturer to buy one controller and use it either way as a SATA SSD controller or as an PCIe flash memory controller. Speeds of each interface indicate this is true at the theoretical throughput end of the scale. LSI reports the PCIe throughput it not too far off the theoretical MAX, (~1.45GB/sec range). Not bad for a chip that can also be use as an SSD controller at 500MB/sec throughput as well. This is going to make designers and hopefully consumers happy as well.
On a more technical note as written about in earlier articles mentioning the great Peak Flash memory density/price limit, LSI is fully aware of the memory architectures and the faillure rates, error rates they accumulate over time.
What would happen if we replaced those 16 disk-based V7000s with all-flash V7000s? Each of the disk-based ones delivered 32,502.7 IOPS. Let’s substitute them with 16 all-flash V7000s, like the one above, and, extrapolate linearly; we would get 1,927,877.4 SPC-1 IOPS – nearly 2 million IOPS. Come on IBM: go for it.
That’s right, IBM is understanding the Flash-based SSD SAN market and is making some benchmark systems to help market its disk arrays. Finally we’re seeing some best case scenarios for these high end throughput monsters. It’s entirely possible to create a 2Million IOPS storage SAN. You just have to assemble the correct components and optimize your storage controllers. What was once a theoretical maximum throughput (1M IOPs) is now achievable without anything more than a purchase order and an account representative from IBM Global Services. It’s not cheap, not by a longshot but your Big Data project or OLAP with Dashboard may just see orders of magnitude increases in speed. It’s all just a matter of money. And probably some tweaking via an IBM consultant as well (touche).
Granted that IBM doesn’t have this as a shipping product isn’t really the point. On paper what can be achieved by mixing matching enterprise storage appliances and disk arrays and software controllers is beyond what any other company is selling IS the point. There’s a goldmine to be had if anyone outside of a high frequency trading skunkworks just shares a little bit of in-house knowledge product familiarity. No doubt it’s not just the network connections that make things faster it is the IOPs that will out no matter what. Write vs. Read and latency will always trump the fastest access to an updated price in my book. But I don’t work for a high-frequency trading skunkworks either, I’m not privy to the demands made upon those engineers and consultants. But still we are now in the best, boldest time yet of nearly too much speed on the storage front. Only thing holding us back is the network access times.
- Extreme Blogging (ibm.com)
- IBM i Storage Options Overview (Which storage is right for me?) (ibm.com)
Does Fusion-io have a sustainable competitive advantage or will it get blown away by a hurricane of other PCIe flash card vendors attacking the market, such as EMC, Intel, Micron, OCZ, TMS, and many others?
More updates on the data center uptake of PCI SSD cards in the form of two big wins from Facebook and Apple. Price/Performance for database applications seems to be skewed heavily to Fusion-io versus the big guns in large scale SAN roll-outs. It seems like due to the smaller scale and faster speed PCI SSD outstrips the resources needed to get an equally fast disk based storage array (including power, and square feet taken up by all the racks). Typically a large rack of spinning disks can be aggregated by using RAID drive controllers and caches to look like a very large high speed hard drive. The Fibre Channel connections add yet another layer of aggregation on top of all that so that you can start splitting the underlying massive disk array into virtual logical drives that fit the storage needs of individual servers and OSes along the way. But to get sufficient speed equal to a Fusion-io style PCI SSD, say to speed up JUST your MySQL server the number of equivalent drives, racks, RAID controllers, caches and Fibre Channel host bus adapters is so large and costs so much, it isn’t worth it.
A single PCI SSD won’t quite have the same total storage capacity as say that larger scale SAN. But for a single, say one-off speed up of a MySQL database you don’t need the massive storage so much as the massive speed up in I/O. And that’s where the PCI SSD comes into play. With the newest PCI 3.0 interfaces and utilizing 8x (eight PCI lane) connectors the current generation of cards is able to maintain 2GB/sec through put on a single PCI card. To achieve that using the older SAN technology is not just cost prohibitive but seriously SPACE prohibitive in all but the largest of data centers. The race now is to see how dense and energy efficient a data center can be constructed. So it comes as no surprise that Facebook and Apple (who are attempting to lower costs all around) are the ones leading this charge of higher density and higher power efficiency as well.
Don’t get me wrong when I tout the PCI SSD so heavily. Disk storage will never go away in my lifetime. It’s just to cost effective and it is fast enough. But for the SysOps in charge of deploying production Apps and hitting performance brick walls, the PCI SSD is going to really save the day. And if nothing else will act as a bridge for most until a better solution can be designed and procured in any given situation. That alone I think would make the cost of trying out a PCI SSD well worth it. Longer term, which vendor will win is still a toss-up. I’m not well versed in the scale of sales into Enterprises of the big vendors in the PCI SSD market. But Fusion-io is doing a great job keeping their name in the press and marketing to some big identifiable names.
But also I give OCZ some credit to with their Z-Drive R5 though it’s not quite considered an Enterprise data center player. Design wise, the OCZ R5 is helping push the state of the art by trying out new controllers, new designs attempting to raise the total number of I/Os and bandwidth on single card. I’ve seen one story so far about a test sample at Computex(Anandtech) that a brand new clean R5 hit nearly 800,000 I/Os in benchmark tests. That peak peformance eventually eroded as the flash chips filled up and fell to around 530,000 I/Os but the trend is clear. We may see 1million IOPs on a single PCI SDD before long. And that my readers is going to be an Andy Grove style 10X difference that brings changes we never thought possible.
- SanDisk Reveals Lightning PCI Express SSD Cards (news.softpedia.com)
- Intel’s SSD 910: Finally a PCIe SSD from Intel (anandtech.com)
- Three questions Fusion-io’s rivals face after flash API bombshell (go.theregister.com)
- Souping Up the Mac Pro: OWC Accelsior PCI Express SSD (barefeats.com)
NoSQL database supplier Couchbase says it is tweaking its key-value storage server to hook into Fusion-ios PCIe flash ioMemory products – caching the hottest data in RAM and storing lukewarm info in flash. Couchbase will use the ioMemory SDK to bypass the host operating systems IO subsystems and buffers to drill straight into the flash cache.
Can you hear it? It’s starting to happen. Can you feel it? The biggest single meme of the last 2 years Big Data/NoSQL is mashing up with PCIe SSDs and in memory databases. What does it mean? One can only guess but the performance gains to be had using a product like CouchBase to overcome the limits of a traditional tables/rows SQL database will be amplified when optimized and paired up with PCIe SSD data stores. I’m imagining something like a 10X boost in data reads/writes on the CouchBase back end. And something more like realtime performance from something that might have been treated previously like a Data Mart/Data warehouse. If the move to use the ioMemory SDK and directFS technology with CouchBase is successful you are going to see some interesting benchmarks and white papers about the performance gains.
What is Violin Memory Inc. doing in this market segment of tiered database caches? Violin is teaming with SAP to create a tiered cache for the HANA in memory databasefrom SAP. The SSD SAN array provided by Violin could be multi-tasked to do other duties (providing a cache to any machine on the SAN network). However, this product most likely would be a dedicated caching store to speed up all operations of a RAM based HANA installation, speeding up Online transaction processing and parallel queries on realtime data. No doubt SAP users could stand to gain a lot if they are already invested heavily into the SAP universe of products. But for the more enterprising, entrepreneurial types I think Fusio-io and Couchbase could help get a legacy free group of developers up and running with equal performance and scale. Which ever one you pick is likely to do the job once it’s been purchased, installed and is up and running in a QA environment.
- Fusion-io Software Development Kit Enables Native Flash Memory Access (sacbee.com)
- Roundup: Fusion-io, Oracle, Teradata (datacenterknowledge.com)
- Fusion-io shoves OS aside, lets apps drill straight into flash – The Register (carpetbomberz.com)
As a result of this impending price war, if you are planning on upgrading your system with an SSD, you might consider waiting for a few months to watch the market and see how much prices fall.
Great analysis and news from Topher Kessler at C|Net regarding competition in the flash memory industry. I have to say keep your eyes peeled between now and September and track those prices closely through both Amazon and Newegg. They are neck and neck when it comes to prices on any of big name brand SSDs. Samsung and Intel would be at the top of my list going into the Fall, but don’t be too quick to purchase your gear. Just wait for it as Intel goes up against OCZ and Crucial and Kingston.
The amount of change in prices will likely vary based on total capacity of each drive (that’s a fixed cost due to the chip count in the device). So don’t expect a 512GB SSD to be dropping by 50% by the end of Summer. It’s not going to be that drastic. But the price premium brought about by the semi-false scarcity of the SSDs is what is really going to be disappearing once the smaller vendors are eliminated from the market. I will be curious to see how Samsung fares in this battle between the other manufacturers as they were not specifically listed as a participant in the price war. However being a chip manufacturer gives them a genuine advantage as they supply many of the people who design and manufacture SSDs with Flash memory chips.
- Low-cost Intel 330 series SSDs sport SandForce SF-2281 SSD Controller (denalimemoryreport.wordpress.com)
- SSD Price Wars (robert.accettura.com)
- SSD Prices To Fall Thanks To Price War (itproportal.com)
Like the native API libraries, directFS is implemented directly on ioMemory, significantly reducing latency by entirely bypassing operating system buffer caches, file system and kernel block I/O layers. Fusion-io directFS will be released as a practical working example of an application running natively on flash to help developers explore the use of Fusion-io APIs.
via (Chris Mellor) Fusion-io shoves OS aside, lets apps drill straight into flash • The Register.
Another interesting announcement from the folks at Fusion-io regarding their brand of PCIe SSD cards. There was a proof of concept project covered previously by Chris Mellor in which Fusion-io attempted to top out at 1 Billion IOPs using a novel architecture where PCIe SSD drives were not treated as storage. In fact the Fusion-io was turned into a memory tier bypassing most of the OSes own buffers and queues for handling a traditional Filesystem. Doing this reaped many benefits in terms of depleting the latency inherent with a FileSystem and how it has to communicate through the OS kernel through to the memory subsystem and back again.
Considering also work done within the last 4 years or more using so-called “in memory’ databases and big data projects in general a product like directFS might pair nicely with them. The limit with in memory databases is always the amount of RAM available and total number of cpu nodes managing those memory subsystems. Tack on the necessary storage to load and snapshot the database over time and you have a very traditional looking database server. However, if you supplement that traditional looking architecture with a tier of storage like the directFS the SAN network becomes a 3rd tier of storage, almost like a tape backup device. Sounds interesting the more I daydream about it.
- Three questions Fusion-io’s rivals face after flash API bombshell (go.theregister.com)
- Fusion-io SDK gives developers native memory access, keys to the NAND realm (engadget.com)
- Fusion-io demos billion IOPS server config – The Register (carpetbomberz.com)
Microsoft and University of California San Diego researchers have said flash has a bleak future because smaller and more densely packed circuits on the chips silicon will make it too slow and unreliable. Enterprise flash cost/bit will stagnate and the cutting edge that is flash will become a blunted blade.
via Flash DOOMED to drive itself off a cliff – boffins • The Register. As reported by Chris Mellor for The Register (http://www.theregister.co.uk/)
More information regarding semiconductor manufacturers rumors and speculation of a wall being hit in the shrinking down of Flash memory chips. (see this link to the previous Carpetbomber article from Dec. 15). This report has a more definitive ring to it as actual data has been collected and projections based on models of that data. The trend according to these researchers is lower performance due to increasingly bad error rates and signaling on the chip itself. Higher Density chips = Lower Performance per memory cell.
To hedge against this dark future for NAND flash memory companies are attempting to develop novel and in cases exotic technology. IBM has “racetrack memory“, Hewlett-Packard and Hynix have MemRistor and the list goes on. Nobody in the industry has any idea what comes next so bets are being placed all over the map. My advice to anyone reading this article is do not choose a winner until it has won. I say this as someone who has watched a number of technologies fight for supremacy in the market. Sony Betamax versus JVC VHS, HD-DVD versus Blu-ray, LCD versus Plasma Display Panel, etc. I will admit at times the time span for these battles can be waged over a longer period of time, and so it can be harder to tell who has won. But it seems to be shorter time spans over the life of these products as more recent battles have been waged. And who is to say, Blu-ray hasn’t really been adopted widely enough to say it is the be all and end all as DVD and CD disks both are used widely as recordable media. Just know that to go any further in improving the cost vs. performance ratio NAND will need to be forsaken to get to the next technological benchmark in high speed, random access, long term, durable storage media.
Things to look out for as the NAND bandwagon slows down are Triple Level Memory cells, or worse yet Quadruple Level cells. These are not going to be the big saviors the average consumer hopes they will be. Performance of Flash memory that gangs up the memory cells also has higher error rates at the beginning and even higher over time. The amount of cells assigned for being ‘over-provisioned’ will be so high as to negate the cost benefit of choosing the higher density memory cells. Also being touted as a way forward to stave off the end of the road are error correcting circuits and digital signal processors onboard the chips and controllers. As the age of the chip begins to affect its reliability, more statistical quality control techniques are applied to offset the losses of signal quality in the chip. This is a technique used today by at least one manufacturer (Intel), but how widely it can be adopted and how successfully is another question altogether. It would seem each memory manufacturer has its own culture and as a result, a technique for fixing the problem. Who ever has the best marketing and sales campaigns will as past history has shown will be the winner.
- AnandTech – Intel and Micron IMFT Announce Worlds First 128Gb 20nm MLC NAND (carpetbomberz.com-Dec. 15, 2011)
- Flash Memory (dmohankumar.wordpress.com)
- The sky is falling! The sky is falling! Paper predicts the bleak future of SSDs and NAND Flash memory (denalimemoryreport.wordpress.com)
Finally theres talk about looking at other interfaces in addition to SATA. Its possible that we may see a PCIe version of SandForces 3rd generation controller.
Some interesting notes about future directions SandForce might take especially now that SandForce has been bought out by LSI. They are hard at work attempting to optimize other parts of their current memory controller technology (speeding up small random reads and writes). There might be another 2X performance gain to be had at least on the SSD front, but more importantly is the PCI Express market. Fusion-io has been the team to beat when it comes to integrating components and moving data across the PCIe interface. Now SandForce is looking to come out with a bona fide PCIe-SSD controller which up until now has been a roll-your own type affair. The engineering and design expertise of companies like Fusion-io were absolutely necessary to get a PCIe SSD card to market. Now that playing field too will be leveled somewhat and possibly now competitors will enter the market with equally good performance numbers
But even more interesting than this wrinkle in the parts design for PCIe SSDs is the announcement earlier this month of Fusion-io’s new software interface for getting around the limits of File I/O on modern day OSes. Auto Commit Memory: “ACM is a software layer which allows developers to send and receive data stored on Fusion-io’s ioDrive cards directly to and from the CPU, rather than relying upon the operating system”(Link to The Verge article listed in my Fusion-io article). SandForce is up against a moving target if they hope to compete more directly with Fusion-io who is now investing in hardware AND software engineering at the same time. 1 Billion IOPS is nothing to sneeze at given the pace of change since SATA SSDs and PCIe SSDs hit the market in quantity.
Everspin on Wednesday said its MRAM magnetoresistive random access memory is trickling into products that require reliable, fast non-volatile memory that can preserve data in the event of a power failure.
Magneto-Resistive RAM in the news
I haven’t heard any product announcements in a while. But it appears Everspin is keeping the faith and shipping real products to real manufacturers. I couldn’t be happier that it’s now on the market and competing for some product designs head to head with RAM and Flash memory. But in this instance it’s really competing against a whole other main stream product; static RAM.
The so-called SRAM was always used as a high speed read mostly cache that allowed a good sized buffer to stay close to the CPU. Static RAM caches were the easiest (but maybe not most cost effective) way to bump the speed of any Motorola or Intel cpu during their co-domination of the desktop market (Intel 386 and Motorola 680000). Stick an SRAM between the CPU and the motherboard, and voila 10-15% performance increase versus a straight through connection between CPU and the motherboard. And static RAM much like Flash based memory chips could also be used to hold data resident for many days powered down. But the cost versus Flash makes it much less competitive. However MRAM can also be used where you might have used a static RAM in the past. Current manufacturers are using it in place of static RAM in hard drive Host Bus Adaptors. This is not just a cost savings but a material savings as these days it is more common to back any mission critical drive electronics with a super-capacitor.
With Magnetic RAMs you can skip including the super capacitor and let the persistence built-in to MRAM do the rest (no need for refreshes or write/re-writes in the background). It makes me wonder if you also went with a super-capacitor to back everything locally and a Magnetic RAM module as well how big a mess that might give them to manage. But from a risk management standing, how much extra or how much less risk would you incur using MRAM plus Super-capacitors in your Disk Controller? I’m sure the cost of manufacture might not warrant the extra effort, but it would still be cool to see a statistical analysis comparing this ‘belt and suspenders’ extravagant setup versus just MRAM or just Super-capacitors.
- New memory models support system design for next year’s introduction of Everspin STT MRAMs (eda360insider.wordpress.com)
- Elpida prototypes 50nm, 64Mbit, 10nsec Resistive RAM (ReRAM). 30nm production slated for 2013 (denalimemoryreport.wordpress.com)
- ReRAM and the memory chip contest (bbc.co.uk)
Fusion-io has achieved a billion IOPS from eight servers in a demonstration at the DEMO Enterprise event
in San Francisco.
The cracking performance needed just eight HP DL370 G6 servers, running Linux 220.127.116.11-45 on two, 6-core Intel processors, 96GB RAM. Each server was fitted with eight 2.4TB ioDrive2 Duo PCIE flash drives; thats 19.2TB of flash per server and 153.6TB of flash in total.
This is in a word, no mean feat. 1 Million IOPS was the target to beat not just 2 years ago for anyone attempting to buy/build their own Flash based storage from the top Enterprise Level manufacturers. So the bar has risen no less than 3 orders of magnitude higher than the top end from 1 year ago. Add to that the magic sauce of bypassing the host OS and using the Flash memory as just an enhanced large memory.
This makes me wonder, how exactly does the Flash memory get used alongside the RAM memory pool?
How do the Applications use the Flash memory, and how does the OS use it?
Those are the details I think that no one else other than Fusion-io can provide as a value-add beyond the PCIe based flash memory modules itself. Instead of hardware being the main differentiator (drive controllers, Single Level Cells, etc.) Fusion-io is using a different path through the OS to the Flash memory. The File I/O system traditionally tied to hard disk storage and more generically ‘storage’ of some kind is being sacrificed. But I understand the logic, design and engineering of bypassing the overhead of the ‘storage’ route and redefining the Flash memory as another form of system memory.
Maybe the old style Von Neumann architecture or Harvard architecture computers are too old school for this new paradigm of a larger tiered memory pool with DRAM and Flash memory modules consisting of the most important parts of the computer. Maybe disk storage could be used as a mere backup of the data held in the Flash memory? Hard to say, and I think Fusion-io is right to hold this info close as they might be able to make this a more general case solution to the I/O problems facing some customers (not just Wall Street type high frequency traders).