The data center industry is rapidly changing. Generative AI is now driving demand for computing power at an exponential rate, with increasing cloud usage and 5G implementing only adding to the burden. On the precipice of this coming AI wave, technology giant Huawei recently released its top 10 trends for data center facilities in 2024.
[Image credit: Huawei]
Key to these, at a conference unveiling the trends, Yao Quan, President of Huawei’s data center facility domain, defined the three characteristics of future data centers – reliable, simplified, and sustainable.
It expects the compound annual growth rate (CAGR) of global AI computing power to exceed 80 percent in the next five years, necessitating a transition from cloud data centers to cloud and intelligent computing data centers.
Of course some of these trends follow on closely from those of last year. In its corresponding white paper of 2023, Huawei rightly emphasized the growing importance of sustainability and low carbon – two key market trends that dominated the industry. This will continue in 2024, but the most recent white paper perhaps goes someway to suggesting how this will be facilitated.
Likewise, the interconnectivity between all these elements is a thread that runs through the most recent trends as well as those of last year. Sustainability, high density racks, cooling solutions, energy efficiency and reliability, all need to operate in unison to gain the best of all of them.
It is this backdrop that underpins the company’s expectations for 2024.
[Image credit: Huawei]
Ensuring Secure and Reliable Data Center Operations
Trend 1: High-reliability products and professional services are the key to ensuring secure and reliable data center operation
According to Huawei, one of the short comings of data centers has historically been “the lack of attention to safety and reliability”. To achieve secure and reliable DC operations, it highlights the need for “full-chain safety” implemented throughout product design and manufacturing.
The company suggests increased use of automation should be used to reduce human interventions, and also highlights the need for appropriate countermeasures after problems arise. Here, professional deployment and services will reduce product failure rates and minimize post-disaster impacts.
Trend 2: Distributed cooling architecture will become a better choice for ensuring safe cooling
Large scale data centers traditionally rely on centralized cooling architecture. This interdependency leaves the whole system, and indeed the potential operation and safety of the facility itself, in a lot of risk – a failure at any single point can impact everything.
Huawei suggests this will drive the trend towards distributed cooling architecture, with independent subsystems that offer more flexibility, and where a small single-point fault will not bring down the whole system.
By decentralizing the cooling infrastructure, data center operators can improve the overall reliability of cooling systems, reduce the likelihood of widespread disruptions, and ensure the continuous operation of critical IT infrastructure.
[Image credit: Huawei]
Trend 3: Predictive maintenance will become a basic feature of data center infrastructure
Prevention is better than cure, but according to Huawei data center maintenance usually takes place after an event has occurred – the cause only discovered after it is too late. Luckily, however, it believes that the advent of the intelligent computing era will greatly shorten response time to such problems.
On an even more positive note, the company believes that the rapid development in AI technologies will expand the scope of predictive maintenance yet further, allowing proactive efforts to stop issues before they arise.
Trend 4: The lifecycle network security protection system will become a shield of data center facilities
As intelligent technologies keep advancing, it seems likely that network attacks will also become more frequent and pervasive. If these attacks can impact a data center’s power or cooling system, the consequences could be disastrous.
This is why Huawei highlights the need for both hardware and software security systems, suggesting software security must be built on what it calls a “lifecycle network security protection system”, encompassing three areas; supply security, in-depth defense, and O&M/operation security.
During data center O&M, the incorrect or missing configuration of configuration items is an important factor leading to vulnerabilities. Any malicious attacks must be detected in real time. Huawei believe an end-to-end control process is an essential cornerstone of network security, helping to establish situational awareness, security configuration, certificate management, and vulnerability management capabilities.
Optimizing Efficiency and Flexibility
Trend 5: Prefabricated and modular solutions will become an optimal choice for high-quality and fast delivery
As demand for new data center continues to grow, the construction of these facilities will need to match the rapid pace the industry is facing. Here, Huawei suggests that a prefabricated, modular approach, will be the quickest and most reliable choice.
With both design and prefabrication taking place early in the factory, these solutions shorten delivery time, reduce onsite construction waste and offer fast service rollout. In turn, this methodology facilitates rapid deployment of new data center builds to meet growing demand, where previously construction time would have been a bottleneck.
Trend 6: Professional Management Platform Makes Data Center O&M More Secure and Efficient
As data centers scale up, so too does the overall complexity for both operations and maintenance. Traditionally it has been difficult to perform all-round inspections, which require highly skilled personnel and can take a long time to locate faults.
This is why Huawei expects a trend of increased use of professional management platforms, reducing this complexity and offering more timely fault location and rectification. The platforms offered by original vendors will help to build in-depth device management capabilities, greatly simplifying O&M.
Trend 7: The convergence of air and liquid cooling becomes the preferred architecture in uncertain service requirement scenarios
Huawei believes the industry stands on the transition from general-purpose computing to intelligent computing, and suggests scenarios supported by both may co-exist in a data center.
The cooling requirements for each differs however, and in general air cooling is sufficient for general-purpose, while liquid cooling will be needed for intelligent computing racks. This, it says, means convergence of the two will become the preferred architecture – offering flexibility and adaptability to future scenarios.
Trend 8: Indirect evaporative cooling is still the best refrigeration scheme now and in the future
Huawei believes “air-cooling solutions are still the undisputed champion of mainstream application scenarios”, making it the most cost effective and most reliable system, and therefore likely to continue the trend going forward.
The company also believes that in response to intelligent computing power demands, this system future-proofs the cooling architecture, allowing it to further adapt to liquid-cooled computing scenarios. As data centers strive to minimize the environmental footprint and optimize resource utilization, the adoption of air-cooling solutions offers tangible benefits.
Driving Sustainability and Energy Efficiency
Trend 9: To further reduce PUE, the optimal solution is to shift the focus from efficient components to system engineering optimization
In order to meet the growing sustainability demands placed on them, data centers have traditionally focused on improving equipment efficiency. Huawei suggests physical limitations mean these improvements are about to hit a bottleneck however, only offering minimal returns on time and investment.
The industry should turn, therefore, to systems engineering optimization in order to balance the actual conditions and component technology level.
Trend 10: AI optimization will become the optimal choice for intelligent optimization of energy efficiency for existing data centers
To meet increased energy-saving requirements, Huawei believe the urgent modernization of data centers is required. In terms of potential downtime and overall effect, it suggests AI energy efficiency optimization will be far superior to traditional manual renovations.
By utilizing preset AI algorithms and big data to optimize the energy efficiency of existing data centers, the solution “facilitates the transition from traditional cooling to intelligent cooling”.