I have just added a new white paper onto the Objective Analysis website: Matching Flash to the Processor – Why Multithreading Needs Parallelized Flash.
This document examines the evolution of today’s CPUs, whose clock frequencies have stopped increasing, but now exploit parallelism to scale performance. Multiple DRAM channels have also been added to performance computing to add parallelism to the memory channel.
Storage hasn’t kept pace with this move to parallelism and that is limiting today’s systems.
New NAND flash DIMMs recently introduced by Diablo, SanDisk, and IBM, provide a reasonable approach to adding parallel flash to a system on the its fastest bus – the memory channel. This white paper shows that storage can be scaled to match the processor’s growing performance by adding flash DIMMs to each of the many DRAM buses in a performance server.
The white paper is downloadable for free from the Objective Analysis home page. Have a look.
On Thursday IBM announced its X6 product family, the sixth generation of the company’s successful EXA server architecture. A smaller byline of the introduction was the company’s new eXFlash memory-channel storage or eXFlash DIMM which is offered as one of many flash options available to X6 users.
Today NAND flash is being shoehorned into HDD formats simply because it is persistent – the data doesn’t disappear when the lights go out. This approach fails to take advantage of NAND’s greatest strength – its low cost relative to DRAM – and this prevents it from fully meeting the needs of most data centers.
Since 2004 NAND has been cheaper than DRAM, and today its price per gigabyte is an order of magnitude lower than that of DRAM. NAND is cheaper and slower than DRAM, and HDD is cheaper and slower than NAND.
A role better suited to NAND flash technology is Continue reading
The following is excerpted from an Objective Analysis Alert e-mailed to our clients on 2 July, 2013:
SanDisk Corporation announced on 2 July 2013 an agreement to acquire SMART Storage Systems, the SSD arm of SMART Modular Technologies, for $307 million in cash and equity. The transaction is expected to close in August, 2013.
SMART has strong SSD technology that allows the company to ship MLC-based SSDs with endurance specifications superior to those of some SLC SSDs. The SSD maker had shipments of about $25M in its most recent quarter.
The SMART acquisition will be the fourth Continue reading
SSD-watchers have expressed some concern over the last few years that SSDs cannot be manufactured using advanced NAND flash process geometries. This is because these parts have lower endurance and a larger number of bit errors than NAND made using less-advanced processes – the tighter the process, the shorter the flash’s life, and the more errors it will have.
Fortunately these concerns seem to be Continue reading
SMART Storage Systems has introduced a new enterprise-class SSD that the company says: “increases the endurance of cMLC Flash to a level that makes SLC drives obsolete.” That’s a pretty hefty claim!
The new Optimus Ultra+ SSD is specified at 100K read IOPS and 60K write IOPS, through its 6Gb/s SAS interface. With capacities ranging from 100-800GB, this SSD supports up to 50 full drive writes per day over its 5-year lifespan, double that of the company’s Optimus Ultra which was introduced in February. That’s quite something for an MLC-based SSD.
SMART has tapped into its Guardian technology to reap SLC benefits from MLC flash through both enhanced external and internal algorithms. Like all other SSD makers and SSD controller makers SMART has focused a lot of attention on error correction, DSP, and other means of correcting errors externally to the flash. The company has also partnered with Continue reading
On Monday December 13 SandForce introduced SSD controllers designed specifically for cloud computing applications.
You might wonder what is so different about cloud applications that they need an SSD controller of their own. SandForce makes some interesting points:
- Cloud applications need low latency
- Cloud computing centers, like client SSDs, need a lot of capacity at a very low price Continue reading
There’s a lot of “Fear, Uncertainty, and Doubt” – FUD – circulating about SSDs and their penchant for failure. NAND flash wears out after a set number of erase/write cycles, a specification known as the flash’s endurance.
While some caution is warranted, a good understanding of how SSDs really behave will help to allay a lot of this concern. Continue reading
The SSD Guy has been asked a number of questions lately about SSDs and RAID. Most of these center around the difference in failure behaviors between SSDs and HDDs – HDDs fail randomly (if ever), while SSDs fail relatively predictably due to wear.
Oddly enough, SSD failures due to wear make them a little friendlier than HDDs. The wear mechanism is managed by the controller in the SSD. SSDs have spare blocks, and the controller manages those blocks, so the controller understands exactly how much wear the SSD has undergone and how much room is left before the SSD will start to have difficulties. Continue reading