Intel is announcing the Intel® Optane™ SSD DC P4800X — the first solution that combines the attributes of memory and storage. The DC P4800X provides an industry-leading combination of high throughput, low latency, ultra-high endurance and high Quality of Service (QoS). Intel’s innovative new solution creates a new data storage tier that eliminates data access bottlenecks.
Intel’s DC P4800X provides application acceleration for fast caching and fast storage, can increase scale per server, and achieves reduced transaction costs for latency-critical workloads. The DC P4800X also allows data centers to deploy bigger and more cost-effective datasets, and to extract new insights from larger pools of memory.
The DC P4800X is engineered to deliver 5-8 x faster performance at low queue depth workloads (as compared to Intel’s DC P3700), providing extremely high throughput levels for single accesses, as well as extremely low latency. NAND-based SSDs are frequently measured at queue depths of 32 for SATA, or 128 for NVMe. To demonstrate its maximum throughput, Intel’s DC P4800X is able to attain as much as 500,000 IOPS / ?2GB/s, at a queue depth of 11. This is truly breakthrough technology that is ideally suited to accelerate enterprise applications to breakthrough performance levels.
NAND-based SSDs, when executing random write operations, require a great deal of background storage media management, which can create significant delays to read operations. Intel’s Optane™ SSD DC P4800X delivers consistent read response times, regardless of the level of write throughput being applied to the drive. The DC P4800X’s read response times remain steady at below <30µs while being subjected to up to 2GB/s of random write throughput.
The DC P4800X is the perfect solution for latency-critical enterprise applications, and provides a high level of predictably fast service. Its 99% read response time is a 60 x improvement over high-endurance NAND SSDs under random write workloads. This equates to a significantly higher QoS level.
Enterprise SSDs require higher levels of endurance, which affects both the life expectancy and cost of those drives. The DCP4800X is geared to environments with high levels of write operations, and is able to withstand intense write traffic that is typically demanded of memory. The DC P4800X features extremely high endurance, yet also with extended life expectancy, making it an excellent choice for write-intensive applications, including high performance computing, online transaction processing, write caching and logging.
Intel’s DC P4800X can be deployed by data centers as either fast storage or cache, or as extended memory. Fast storage or fast cache is tiering and layering that enables better memory-to-storage hierarchy. Intel’s DC P4800X represents a new storage tier that eliminates the bottlenecks of typical NAND storage to accelerate applications, and allows each server to produce more work. In the extended memory use case, the DC P4800X joins in a shared memory pool with DRAM at either the OS level or application level to create bigger memory and/or more affordable memory. With bigger memory, larger working datasets can be deployed. This allows for new insights into growing dataset segments that include scientific computing, autonomous driving and healthcare. More affordable memory means that Intel Optane™ SSDs can displace a portion of system DRAM.
Intel’s Optane™ Technology is an innovative combination of 3D XPoint™ memory and Intel’s advanced system memory controller, interface hardware and software IP. The DC P4800x is initially offered in a capacity of 375GB, with larger capacities to follow in subsequent quarters of 2017. The initial version if a half-height, half-length, low-profile add-in card, with U.2 form factor versions to be offered in subsequent quarters of 2017. Its interface is PCIe 3.0 x4 with NVMe protocol. Endurance is stated as 30 drive writes per day (DWPD), or 12.3 Petabytes Written (PBW).
Several pertinent questions were raised during press pre-briefings, and the answers reveal some interesting aspects of how very different the 3D XPoint™ memory-based DC P4800X is from typical NAND-based SSDs.
Q: How much overprovisioning in an Intel Optane™ SSD and how many die, or channels in the SSD? What is the spare area in an Intel Optane™ SSD for? Can you increase the performance of an Intel Optane™ SSD with overprovisioning?
A: An Intel Optane™ SSD is not “overprovisioned” in the same sense as a NAND-based SSD. The random write performance of a NAND-based SSD can be increased by increasing the spare capacity percentage, as this makes the equivalent of defragmentation more efficient. Since 3D XPoint™ memory media is a write-in-place media there is no defragmentation needed, and therefore no performance can be gained in an Intel Optane™ SSD by increasing spare capacity. With that said, both NAND-based and Intel Optane™ SSDs must store ECC and metadata to maintain low error rates throughout the life of the drive, and meet the demanding needs of enterprise customers.
Also, there are significant differences in the 3D Xpoint™ memory media and NAND media. For example, NAND media includes significant extra capacity inside every die for sparing out blocks, as well as managing ECC code words. 3D Xpoint™ memory media does not have such extra capacity. Given these media architectural elements, the resulting SSD architecture and “overprovisioning” cannot be considered in the same way one would consider “overprovisioing” or spare capacity for a NAND-based SSD.
Here are the specifics of the Intel Optane™ SSD DC P4800X: The controller is a 7-channel controller, and performance is best with an even die-to-channel loading. For the 375GB drive, there are 4 media per channel, or 28 die in total. Spare capacity beyond the user capacity is used for ECC, firmware, and ensuring reliability meets the high standards customers expect from an Intel data center SSD.
How is data residing in the RAM Pool protected in the event of a power fail? I suspect it’s not. Anytime external memory is used to boost SSD performance you’ll have this risk.
This is a completely new type of memory. No memory refresh cycles are needed as It doesn’t get erased when it loses power. But if your using it as RAM then why would you store data there? There is no reason to use any kind of write cache with this type of memory/storage