This site uses cookies. By continuing to browse the site you are agreeing to our use of cookies. Read our privacy policy>Search

Reminder

To have a better experience, please upgrade your IE browser.

upgrade
If you need help, please click here:
MENU
Ever Fast

No.1 Performance
with Chip-Powered Architectures

20 Million IOPS* and 0.1 ms Ultra-Low Latency

OceanStor Dorado benefits from five built-in proprietary Huawei intelligent chips to accelerate critical paths for transmission, compute, intelligence, storage, and management. Best-in-class dual controllers performance with 640k IOPS and 0.1ms ultra-low latency.

  • 1
    intelligence
    Ascend 310 AI chip

    Enhance read cache hit ratio by 50%

  • 2
    transmission
    Interface chip

    Shorten read latency by 50%

  • 3
    compute
    Kunpeng 920 chip

    Improve computing power by 25%

  • 4
    management
    BMC Management chip

    93% fault location accuracy.

  • 5
    storage
    SSD controller chip

    Shorten write latency by 50%

  • Ascend AI Chip

    Based on machine learning frameworks, the intelligent AI chip actively analyzes and understands I/O rules of multiple application models, implements intelligent prefetch of memory space, improves the read cache hit ratio by 50%, and shortens batch processing latency from 300μs to 150μs.

    50%

    higher read cache hit ratio

  • Intelligent
    Multi-protocol Interface Chip

    The chip intelligently parses front-end protocols – including 32/16/8 Gbit/s Fiber Channel, 100/40/25/10 Gbit/s Ethernet, with forward compatibility – shorten read latency from 160μs to 80μs by 50%.

    50%

    Read latency

  • Kunpeng 920 Chip

    The intelligent processor chip uses an ARM-based CPU that offers the industry's highest performance: 25% more computing power than the next-best.

    25%

    higher than the industry benchmark

  • BMC Management Chip

    Implementing intelligent and comprehensive fault management, the chip provides fault location accuracy of up to 93%, slashing fault recovery times from two hours to just ten minutes.

    93%

    fault location accuracy

  • SSD Controller Chip

    The chip implements the Flash Translation Layer (FTL) in silicon to accelerate data access in SSDs, reducing read latency from 40μs to 20μs in low-load scenarios.

    20 μs

    read latency

Uncompromising
Performance in Different Scenarios

Competitor vs. Huawei storage performance after value-added features are successively enabled

Competitor AFA: 40% lower; Huawei AFA: <10% lower

  • Test devices: Dual-controller, 1 TB cache, 16G FC frontend, NVMe backend, 25 x 3.84 TB SSDs
  • Test conditions: Hybrid workload, 8K I/Os, 7:3 read/write, 1 ms average latency, 8 LUNs

The Fast Path: NVMe Architecture

Huawei makes the most of NVMe protocols with the industry’s only end-to-end development of an NVMe flash controller, NVMe all-flash Operating System (OS), and NVMe SSDs. Powered by this NVMe architecture, with finely tuned disk/controller coordination, Huawei’s NVMe all-flash storage now offers stable latency of 0.1 milliseconds.

  • 0.1ms
    Ultra-Low Latency

    Test mode: 7:3 Mixed read/write, 130μs read, 50μs write Total latency: 130*0.7 + 50*0.3 =106μs