Data awakening: building leading AI-ready data infrastructure

The 2024 Huawei Innovative Data Infrastructure (IDI) Forum took place in Sandton, Johannesburg recently and Diego Xiao, Vice President of Huawei Data Storage Product Line, delivered the day’s keynote address on “Data Awakening: Building Leading AI-Ready Data Infrastructure”. Xiao introduced the “1+3+X” architecture for AI-ready data infrastructure, which stands for: one data lake, three data pools, and X data services. This architecture will help enterprises fully unleash the value of their data and facilitate the intelligent upgrade of Sub-Saharan Africa Africa’s digital economy. 

Diego Xiao, Vice President of Huawei Data Storage Product Line

Recent years have seen the innovation and commercialisation of new information technologies with a focus on generative AI. Large AI models are now widely used for mission-critical services in industries such as finance, public services, and carriers.  Xiao shared the success story of a leading bank in Asia that uses Huawei’s OceanStor storage solution for their AI system. Their call centre’s average handle time is now 20 percent shorter; potentially fraudulent transactions are now flagged within 10 milliseconds, enabling real-time fraud protection; the loan application process has been shortened from 15 minutes to just one minute and approval can be accomplished in just one second.

Xiao says, this example shows that efficient management and use of data assets now present the greatest opportunity and challenge for enterprises in the AI age. To capitalise on this businesses, need to build their data infrastructure using a unified data lake and three data pools that deliver exceptional performance, large capacity, and rigorous data protection. This kind of infrastructure meets the requirements for applications in terms of performance, reliability, scalability, and resilience. Huawei’s AI-ready data infrastructure can help enterprises develop a competitive edge in the AI age.

Huawei OceanStor Dorado all-flash storage and OceanStor A800 high-performance storage establish a high-performance data pool to support mission-critical services and AI applications

Huawei OceanStor Dorado all-flash storage series delivers ever-fast, ever-solid, and innovative services. Its unique gateway-free, active-active SAN/NAS architecture and 99.99999 percent reliability ensure core service continuity 24/7. The FlashLink disk-controller collaboration algorithm boosts file access performance fivefold. In addition, the container ecosystem is supported by plugins (CSM, CSI, CDR, and more) which enable the agile development and deployment of applications for enterprises.

The OceanStor A800 high-performance storage, a member of the Huawei OceanStor A series, is designed for AI. It delivers four times higher bandwidth and eight times higher IOPS than leading alternatives. It supports EB-scale capacity expansion as well as up to 4 096 computing cards and 512 controllers. In addition, the built-in vector index, tensor data, and RAG knowledge repositories can power AI services in a vast range of industries. 

Huawei OceanStor Pacific scale-out storage underpins the large-capacity pool and activates mass unstructured data

The rise of new technologies and applications such as big data, HPC, and AI has generated massive amounts of unstructured data. Huawei OceanStor Pacific’s scale-out storage series supports data interworking between different protocols such as NFS, HDFS, and S3, eliminating the need for data migration. It contains up to 36 disks (up to 61.44 TB per disk) in a 2 U chassis and offers up to 1 PB/U capacity density—the highest in the industry—while only consuming 0.5 W/TB. It also supports multi-active setup across 3 to 12 sites with zero RPO.

Huawei OceanProtect appliance series establishes a solid data protection pool to comprehensively safeguard enterprises’ data assets

Data resilience and privacy protection have become increasingly important in the AI age. Enterprises could suffer catastrophic losses from cyber-attacks that encrypt, steal, or tamper with mission-critical assets such as model files and training data. Huawei OceanProtect X and E series appliances provide end-to-end data protection, with fast backup and recovery, large capacity, and space savings. Both offer outstanding physical bandwidths for backup (155 TB/h) and recovery (172 TB/h). In addition, the patented deduplication and compression algorithm delivers an industry-leading data reduction ratio that is 65 percent higher than leading alternatives. This helps customers reduce their total cost of ownership by 30 percent.

Omni-Dataverse global file system builds a physically distributed and logically unified enterprise data lake

To be AI-ready, you need to get data-ready. The Omni-Dataverse global file system, which is housed in the DME, manages the metadata on Huawei storage devices across different data centres in a unified manner. It builds a complete data lake and provides a global view of data assets. Interfaces on storage devices can be invoked to control data flow, making enterprise data assets visible, manageable, and mobile across regions.

DataMaster O&M model uses AI to enhance enterprise O&M

At this forum, Huawei released DataMaster, a large AI model for IT O&M. DataMaster analyses O&M logs and quickly identifies irregularities and nips issues in the bud. It can understand user intentions, translate them into SQL statements, retrieve data from the database, and flexibly generate reports as needed. It also converts an engineer’s instructions into machine commands and efficiently executes O&M operations by invoking various tools or APIs. DataMaster will be integrated into the DME Data Centre Management platform, which is expected to revolutionise data infrastructure management and O&M in the future.

DCS full-stack data centre solution provides diverse data services to unleash the value of data assets

Making the most of data by converting it into valuable assets is a top priority for many enterprises. Huawei’s DCS full-stack data centre solution uses AI technologies to deliver comprehensive infrastructure as a service (IaaS) and platform as a service (PaaS). These include virtualisation and containers, big data analytics, an AI development platform, and data backup. With this Huawei aims to help customers build an efficient, resilient, reliable, innovative, and future-oriented data centre foundation.

Diego Xiao ended by emphasising that Huawei remains committed to meeting the needs and addressing challenges faced by customers in the era of data awakening, with the goal of building AI-ready data infrastructure and unleashing the infinite potential of digitalisation and intelligence in Sub-Saharan Africa.


About Author

Related News