Building green financial data centers on 4 key storage technologies



[Guizhou, China, May 26, 2022] At the China International Big Data Industry Expo under the theme Eastern Data, Western Computing (Cross-Region Data Processing): Building a National Computing Network System, Huawei and related organizations jointly released the research report on the development of data centers. ecological and high-quality data for peak carbon and carbon neutrality (hereinafter referred to as “the research report”).

More importantly, the research report focuses on the following four storage technology directions in the discussion of technical trends and solutions to promote green data centers (DCs), especially for the financial sector.

Direction 1: Replace HDDs with SSDs to Prepare for the Flash-Ready DC Age

Flash media exhibits high density and reliability with low latency and low power consumption rates. By providing the capacity of hard drives, flash storage reduces power consumption by 70%, boosts performance by 100x and reduces footprint by 50%. All-flash is the inevitable future trend for data center storage media. The percentage of flash storage media in data centers is gradually increasing. Organizations that want to stay ahead of the game are recommended to deploy 50% all-flash by 2025, 90% by 2030, before eventually operating 100% all-flash data centers.

Figure 1. All-flash storage reduces carbon footprint by 50%

Orientation 2: Accelerate the adoption of the decoupled storage-computing architecture for Big Data

Typically, a general purpose server is equipped with a limited number of hard drives. In contrast, a purpose-built high-density storage node provides 2 to 2.6 times more dense hardware, with 10% to 30% lower power consumption at equivalent capacity. For example, when analyzing big data, the decoupled storage and compute architecture combines with erasure coding (EC) technology to improve disk utilization by 33% to 91%, reducing occupied disk space and significantly reducing power consumption.

Figure 2. Decoupled Compute and Storage Architecture Reduces Power Consumption by 10-30%

Direction 3: Improve data storage efficiency with data deduplication and compression

Data deduplication and compression technologies run a series of state-of-the-art algorithms to improve data storage layout and efficiency. These technologies are already well known in the industry with flash media and are used to improve the data reduction ratio in a range of 2:1 to 3.6:1 in database, desktop cloud scenarios and virtual machine, while reducing energy consumption by at least 50%.

Figure 3. Super Deduplication Increases Data Reduction Rate by 25%

Direction 4: Advance data-intensive clustered storage technologies

Data-intensive clustered storage refers to an integrated solution of storage, computing, networking, and power that helps build green midsize and large data centers. Innovative technologies such as high-density and high-ratio EC, data processing unit (DPU) offloading, data processing acceleration, and storage liquid cooling are essential to ensure high delivery efficiency and rapid service deployment of DCs. Additionally, one pool connects to multiple clouds to implement unified and converged storage resource pools across data centers, improving resource utilization by 50%.

Figure 4. Data-intensive clustered storage reduces total cost of ownership by 30-50%

Building a green and smart financial data center places higher demands on data center design, power utilization efficiency (PUE), and IT equipment. By providing a rich portfolio of all-flash domain controllers and data-intensive storage, Huawei is accelerating the upgrade of the storage industry and leading the world towards green and smart data centers.


Comments are closed.