Enterprise datacenters were traditionally manually operated and maintained, demanding both constantly available personnel and persistent vigilance for potential maintenance needs. But AI advancements are changing that status quo, IDC notes in a report backed by Chinese networking giant Huawei, and moving towards datacenter automation analogous to autonomous driving: five levels of increasingly computer-controlled operations and maintenance that will eventually see intelligent systems handle everything from problem detection to resolution based on human intent and experience.
According to IDC, enterprises across multiple industries are struggling to guarantee application and data resilience — an issue enhanced by the COVID-19 pandemic — as the datacenter network has become “the digital nervous system” for newly far-flung businesses. Consequently, enterprises are now actively planning to automate their datacenters; between 20-30% of surveyed businesses said full autonomy is a goal today, while over 90% are planning for it within a multi-year time frame.
The IDC report is significant for technical decision makers because it simultaneously identifies an issue of increasing importance to modern enterprises and provides a straightforward roadmap to incrementally resolving that issue. IDC and Huawei suggest that moving from manual datacenter operations to full autonomy will require five steps paralleling automotive AI, gradually shifting solution execution, problem awareness, analysis, immediate decisions, and big picture policy choices from people to AI systems, eventually ceding control in all datacenter operation and maintenance scenarios rather than just select ones.
Huawei is pitching CloudFabric as the first solution with “L3 autonomous driving” for datacenters, noting that it uses AI and machine learning to achieve conditional network autonomy — a stage where software handles datacenter problem awareness and solution execution, but partially relies upon humans for analysis and decision making, and fully relies upon them for intent and experience-based policy choices. CloudFabric, it notes, can self-heal datacenter issues and self-optimize networks, and is already in use by over 2,800 enterprise datacenters “globally.”
But while IDC notes that organizations at any level of datacenter autonomy can move forward to the next stage, and stands ready to help, it conspicuously recommends that enterprises “leverage assistance from trusted vendors and third parties.” Unfortunately for Huawei, trust has been clouding its business for the past two years. Despite protestations from Huawei, the U.S. government and allies across Europe and Asia have deemed the company’s networking gear untrustworthy, requiring carriers to rip out and replace Huawei hardware with alternatives from trusted vendors. Governments and businesses alike are concerned that Huawei-managed data could be harvested or manipulated by the Chinese government, which has reserved the right to demand cooperation in surveillance activities from domestic vendors.
The “global” list of CloudFabric enterprise datacenter users is top lined by largely Chinese companies, including Tencent, Baidu, China’s State Grid, and multiple Chinese banks, while Sony’s name is a standout from Japan. Arista, Cisco, and HPE are among Huawei’s competitors in the space.