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

If you need help, please click here:

Creating Data Storage with Superb Cost-Effectiveness per Bit

2021-09-27 198
Emotional male voice
  • Emotional male voice
  • Emotional female voice
  • Standard male voice
  • Standard female voice
Speed 1.0X
  • Speed 0.5X
  • Speed 1.0X
  • Speed 1.5X

It aims to build with memory and data as the center of the new data storage system based on innovative software and hardware technologies, such as self-driven storage system, transparent and secure large memory, data protection in massive data scenarios, data computing power network, storage system framework with separated storage and compute, and storage system oriented to the emerging data ecosystem, achieving superb cost-effectiveness per bit.

Challenge Direction 1: Warm and Cold Storage System Technology with the Highest Energy Efficiency per Bit

1. Low-cost cold storage system technology oriented to high-density optical storage: next-generation cold storage system better than the existing systems’ TCO.

2. Superb cost-effective large-capacity warm storage system technology: large-ratio data redundancy algorithm; warm-storage-oriented data encoding format and reduction algorithm for improving space utilization; low-energy-consumption data access; and unified storage access protocol and related instruction-sets of multiple media

3. Data protection technology for new massive data scenarios: non-intrusive CDC, large-scale system backup, trusted data protection based on blockchain, source-end data reduction, and trusted CDM for edge, data-center, and multi-cloud scenarios.

4. High-reliability storage technology across multiple data centers: multi-AZ high-reliability erasure coding technology

Challenge Direction 2: Self-Driven Storage System Technology with the Highest Efficiency per Bit

1. Best TCO multi-mode hybrid storage technology of self-driven: Superb cost-effective tiered storage of multiple media.

2. Building storage technology with optimal AI-Native efficiency: AI tuning in all domains (end-to-end I/O paths, system scheduling, fault diagnosis, and analysis and planning), lightweight AI algorithm framework oriented to storage systems, AI Native storage data structure and layout technologies (machine learning index, intelligent tiering RAID, and intelligent GC-free layout)

3. Intelligent O&M technology: intelligent fault demarcation and locating, root cause analysis, application-oriented and learning-based intelligent recommendation O&M, and optimal service placement and networking planning of the multi-service convergence system.

Challenge Direction 3: Key Technology of PB-Level Memory Storage System with the Highest Cost-Effectiveness per Bit

1. Key technology for large-capacity DRAM: Algorithms and OS technologies are used to achieve large-capacity DRAM reliability and memory reduction.

2. Low-cost memory tiering technology: Transparent memory access is provided by using media such as Fast NAND and PCM. Data layout oriented to heterogeneous memory media and adaptive algorithms for data prefetching are used.

3. Key technologies for transparent and secure large memory: The distributed memory pool supports multi-write transactions, multi-tenant security isolation, and data leakage prevention. The long-distance transparent large memory network protocol and distributed large memory virtual memory management system provide large memory access interfaces for database, big data, AI, and HPC scenarios.

4. New native memory application ecosystem: Transparent large memory technology is combined with applications such as databases, graph computing, and real-time transactions.

Challenge Direction 4: Framework of the Storage-Compute Separation Storage System with Optimal Efficiency

1. Storage client technology based on heterogeneous computing power acceleration: heterogeneous computing power acceleration and lightweight protocol access

2. Large-scale RDMA network technology: direct access of 100,000-level clients to the storage resource pool, concurrency control protocol in large-scale networking, and RDMA congestion control algorithm

3. Data computing power network technology: network-storage-compute integrated technology and Computational Storage

4. NDP: a unified NDP framework oriented to scenarios such as databases, big data, AI, and HPC

Challenge Direction 5: Building New Storage Systems for Emerging Data Ecosystems Such as Distributed Databases, Big Data, HPC, and AI

Technology for building new storage systems for applications such as databases, big data, AI, and HPC: database storage engine offload, storage that supports multi-write and multi-read of databases, and schema-aware storage


Write your comment here.


    More comments

      You have scored successfully.

      You have submitted successfully.

      Evaluation failed.

      Submission failed.

      Please write your comment first.