Blog: Huawei Takes Artificial Intelligence to the Database – SDxCentral
Huawei released an artificial intelligence (AI)-native database management product, GaussDB, as part of its greater into automating the network stack. This puts the equipment vendor in competition with traditional database vendors including IBM, Oracle, and Microsoft — some of which also have autonomous database products. The vendor also injected AI into its distributed storage platform, FusionStorage, among other updates.
GaussDB embeds AI into the lifecycle of distributed databases so that organizations can autonomously perform observations and measurements, tuning, diagnosis, and healing. The Chinese equipment vendor claims that the database will improve tuning performance by 60% via its self-tuning algorithm. That level of improvement is applicable to online analytical processing, online transaction processing, and hybrid transaction and analytical processing scenarios, according to Huawei.
The database will run on public and private clouds, but when it runs of Huawei’s cloud the database will be able to provide additional data warehouse services across multiple industries. The computing framework can also run on x86, ARM, graphics processing units (GPU), and network processor units (NPU).
Huawei AI Stack
Huawei has been promoting its work in AI since it launch a series of AI chips called Ascend. In December at its Huawei Connect conference in Shanghai, the vendor announced a full AI-stack of products and cloud services that would be built off of the Ascend chip. The following month it announced a new data center switch that uses its AI chip to boost network performance.
The full AI-stack includes services that deliver abundant and affordable computing power, an AI platform with full-pipeline services, and an adaptive portfolio for cloud, edge, and device environments, according to the company.
Huawei is also investing in AI research and talent as it injects its current portfolio with the technology.
The company says that it is leveraging “AI for good” — identifying ten areas where the technology could enact positive change. These areas include faster model training, abundant and affordable computing power, AI deployment and user privacy, new algorithms, AI automation, practical application, real-time and closed-loop systems, multi-tech synergy, platform support, and talent availability.
This message of enacting positive change comes as the vendor continues to battle the mounting international pressure surrounding the safety and security of its 5G and telecom equipment. The company is accused of putting backdoors in its gear to enable spying on networks that use its equipment, which Huawei has vehemently denied.
Many vendors are putting their money behind AI, particularly chipmakers including Qualcomm, Nvidia, and Intel (in partnership with Facebook), which have all announced some form of an AI chip. Meanwhile, cloud providers Amazon Web Services (AWS) and Google Cloud have begun work on AI stacks and chips for inference workloads.
IBM recently announced its vision for an AI database for cognitive applications that will be primarily suited for AI application developers and data sciences. Oracle has been touting its own “fully-autonomous, self-driving database” for some time. The service is available in the Oracle Cloud with autonomous analytics, mobility, application development, and integration services coming soon.
Huawei also released version 8.0 of its distributed storage platform. The FusionStorage portfolio contains Huawei’s software-defined storage (SDS) products and the platform also provides object storage and unified storage management for unstructured data.
The updates boosted the platform’s read-write performance per-node; support for block, file, object, and Hadoop Distributed File System protocols; and integrated AI into storage management.
By adding AI components into FusionStorage 8.0, the service supports resource planning, service provisioning, system optimization, risk prediction, and fault location.