For laptop owners, flash-memory drives boost battery life and performance while making notebooks lighter and more bearable for frequent business travelers. In the data center, benefits include higher reliability than their magnetic counterparts, lower cooling requirements and better performance for applications that require random access such as e-mail servers.
So far, the biggest barriers to adopting solid-state drives (SSD) in the data center have been price and capacity. Hard disk drives (HDD) are much less expensive and hold much more information. For example, a server-based HDD costs just $1 to $2 per gigabyte, while SSD costs from $15 to $90 per gigabyte, according to IDC.
Capacities are just as disparate. The Samsung SSD drive only holds 64GB, although the company plans to release a new 128GB version next year. Meanwhile, Hitachi America Ltd. makes a 1TB HDD that's energy efficient and priced at $399 for mass deployment in servers.
Enterprise Strategy Group Inc. analyst Mark D. Peters explains that solid-state technology has been on the radar for years, but has not been a "slam-dunk" in terms of price and performance for corporate managers. That's about to change, he says, because the IOPS (input/output operations per second) benefits to SSDs are too impressive to ignore. Advantages include how SSD has no moving parts, lasts longer, runs faster and is more energy efficient than an HDD.
And prices are falling fast. Right now, the industry trend is a 40 percent to 50 percent drop in SSD pricing per year, according to Samsung.
The arrival of hybrid drives such as Samsung's ReadyDrives -- which use both SSD and HDD technology -- and SSD-only servers "suggests the time for SSD as a genuine -- and growing -- viable option is getting closer," says Peters. He was referring to the recent IBM announcement about BladeCenter servers that use a SSD.
"Price erosion, coupled with increased capacity points, will make SSDs an increasingly attractive alternative to HDDs" in data centers, agrees Jeff Janukowicz, an analyst at IDC in Framingham, Mass.
Two examples of how SSDs solve persistent throughput problems for high-performance computing shows how SSD technology may make new inroads in corporations in 2008, some industry watchers believe.
Solid-state at the Stanford Linear Accelerator Center
At this research center, SSD is being used for some of the most data-intensive work going on today. The Stanford Linear Accelerator Center (SLAC) in Menlo Park, Calif., uses particle accelerators to study questions, including where antimatter went in the early universe and what role neurexin and neuroligin proteins play in autism.
The amount of data is immense -- in the petabytes -- and the lab uses a cluster of 5,000 processor cores. Despite that, the discrete chunks of data that are requested and analysed by several hundred researchers are highly granular -- usually just 100 to 3,000 bytes of information. At the same time, scientists tend to perform thousands of data requests, accessing a few million chunks of data per second.
Richard Mount, SLAC's director of computing, explains that the response time for these researchers' data requests is limited not by the number of processors or by the amount of network bandwidth, but rather by disk access time. "Flash memory is over a thousand times faster than disk" drive technology," says Mount. "Hard disks are limited to around 2,000 sparse or random accesses per second. When accessing thousand-byte chunks, this means that a disk can use only 1/50th of a gigabit-per-second network link and less than 1/100,000th of a typical computer center network switch capacity."
This limitation has translated into the need to make what the lab calls "skim data sets." In other words, preassembled collections of related data that at least one researcher has already requested. "There is no waiting for skim data sets that already exist, but if somebody wants one that does not already exist, then they normally have to wait for a skim production cycle that takes place once every four to six months," Mount says.
To help researchers receive data in a more ad hoc manner, flash storage may be just the thing. "We have no religious attachment to flash, but we can construct flash-based storage at a reasonable cost and around 25ms latency, and we are doing so."
SLAC has developed its own SSD-based system that is in the final debugging stages, Mount explains.
"The first version of this will provide about 2TB of storage, but we can easily grow this to 5 or 10TB just by buying flash chips," though he reckons the scalability will require "more serious expenditure." At the 2TB level, it will serve as a test and development system only.
Eventually, the goal is to use SSD technology as a cache for all particle accelerator research, which will allow scientists to access data at any time from any data store. "SSDs help the entire system run more efficiently by ensuring the I/O capability is in balance with the rest of the application system," adds IDC's Janukowicz. "The characteristics of flash-based SSDs make them a well-suited alternative for high-IOPS applications that are read-intensive. SSDs have no rotational latency and have high random-read performance. Thus, with SSDs the time to access the data is consistent and very small regardless of where on the device the data is held."
Considering SSD at the Pacific Northwest National Laboratory
At the Pacific Northwest National Laboratory (PNNL) in Richland, Wash., solid-state technology could help alleviate a supercomputer bottleneck. At the lab, researchers run tests that sustain a write speed of 80Gbit/s and a read speed of 136Gbit/s. Yet, one or two slow hard disk drives running at one quarter the speed of other disks causes performance to degrade quickly.
"Solid-state devices such as flash drives can use a RAID striping technique to achieve high streaming bandwidth -- just like [hard] disk drives -- while also maintaining very low latency for random access," says Robert Farber, a senior researcher at PNNL. "This is a very exciting combination."
The lab has not moved to solid-state technology yet. But Farber says the real debate is whether low-latency access for "seek-limited applications" -- in other words, many requests for small amounts of data -- can alleviate the pressure of computing bandwidth. It is not solely a price-per-gigabyte debate. "It remains to be seen how much of a price premium consumers will tolerate before robustness, power, storage capacity and physical space differences cause a mass departure from magnetic media," Farber says.
At the PNNL, the latency goal for its last supercomputer was 25Mbit/s, per gigaflop of peak rate floating-point performance. This is mostly to be able to handle the data-intensive nature of the NWChem scientific software calculations running. The lab's new environmental molecular sciences facility contains a new supercomputer with a theoretical peak floating point performance of 163 teraflops. And, like at the Stanford lab, disk speed is a critical part of the equation, so solid-state is the forerunner in solving the bottleneck.
One breakthrough Farber expects in the not-too-distant future: Operating systems will change their memory hierarchy to directly access SSD, turning the technology into a hard drive replacement for mass storage.
Complementary, not replacement tech for most users
One question that remains: When will SSD really impact the corporate world? Some say SSD in the data center is just on the horizon, since laptops such as the Dell XPS M1330 uses a Samsung 64GB SSD. Alienware also offers a 64GB option in some of its desktop computers. And SSD is applicable across the commercial landscape; while researchers need the speed to study proteins, retailers may need or want faster POS transactions.
One company to watch in this space: Violin Memory Inc. in Iselin, N.J. The company's Terabyte-Scale Memory Appliance provides over 1Gbit/sec. access for sequential and random-access. SLAC's Mount says he tested a DRAM-based prototype appliance from Violin, and that its upcoming flash-based system "seems a good match for our applications."
A Violin spokesman explains that the two key bottlenecks in corporate computing are network speeds and IOPS for storage systems. Today, disks run at about 100Mbit/s for sequential operations, but only 1Mbit/s for random 4k blocks, he says.
"In some cases, there are minimal capacity requirements which are well suited for SSDs," Janukowicz adds. "Also, in high-performance applications, the IOPS metrics can favor SSDs over HDDs." However, even with all those benefits, he says that "IDC does not see SSDs completely replacing HDDs in servers. SSDs do offer performance advantages and are a 'green' solution. However, there are many applications that require the capacity provided by HDDs."
Enterprise Strategy Group's Peters says that throughput requirements will lead to a gradual shift away from hard disk drives to solid-state technology, but it will take time in the corporate world. "Moving wholeheartedly from one technology to another is a rare thing within data centers," he says.
John Brandon worked in IT management for 10 years before starting a full-time writing career. He lives in Fergus Falls, Minn. and can be reached at [email protected]