Collapse & Return to Top
Purpose-built for the demanding hyperscale and mega datacenters, the Seagate 1.86TB Nytro XP6210 Flash Accelerator Card delivers accelerated performance for read-intensive applications, optimized power and thermals, and an overall lower cost per gigabyte PCIe flash solution. By plugging flash directly into the server’s PCIe slot, these cards can help improve performance and reduce the physical space needed to support high-density enterprise applications. Hyperscale datacenters require cost-effective scaling to thousands of servers to meet the compute and storage demands of Web 2.0, cloud computing and big data analytics applications. These applications, which search data for content or comb data for trends, are primarily reading data from storage. The Nytro XP6210 is designed to reduce latency and delivers high performance by connecting flash closer to the CPU. Multiple Nytro cards can be installed in the same server to get even higher application performance.
Delivering leading endurance and reliability using low-cost flash, this design is optimized for low power and thermal characteristics. The Nytro XP6210 requires up to 30 percent less power than similar cards using MLC flash technology. Unlike other cards with flash memory, this efficient design does not require heat sinks to control temperature. The lower power and cooling translates into lower OpEx. The Nytro XP6210 requires minimal resources
from the CPU and host memory when compared with other cards. It offloads storage management tasks and uses enterprise hardened, standard drivers typically in-box with most operating systems.
Nytro uses SandForce flash controllers and is designed to deliver consistently high levels of performance, endurance, and reliability under demanding workload conditions. Seagate DuraWrite technology optimizes the number of program cycles to the flash storage, effectively extending its rated write endurance by 8 times or more, depending on the use case, when compared to standard controllers for compressible data.