The next instalment of 2024 tech predictions continues with D-F. Read predictions from 5 to C at https://www.techtradeasia.com/2024/01/a-to-z-of-tech-predictions-in-2024-5-c.html.
D is for data and data centres
AI
 |
| Source: Digital Realty. Sharp. |
"They will have to flex to respond to these changing demands – being agile, adaptable, and providing customers with options that allow them to scale or shift direction to make the most of the opportunities AI offers, will be critical,” said Chris Sharp, CTO, Digital Realty.
"The emergence of technologies like AI is opening new doors for optimisation and automation that will deliver better cost control and quality for customers. These advances represent a major transformation that is redefining how data centres operate and serve customers," added Pieter Kraan, MD, Leaseweb Singapore.
Network infrastructure
 |
| Source: CommScope. Peluffo. |
"Enterprises looking to maximise network investments in 2024 will have to evaluate and integrate what will be the most flexible, optimal portfolio spanning on-prem, collocated and edge data centres as well as wired, Wi-Fi, 4G/5G, Internet of Things (IoT) technologies to suit their evolving needs."
Server infrastructure
 |
| Source: OVHcloud. Maiolo. |
"Bare metal servers offer dedicated and single-tenant hardware, ensuring enhanced performance, predictability, and security. This bespoke approach not only addresses the specific computational needs associated with emerging technologies but also caters to the heightened sensitivity and regulatory considerations surrounding the processing of confidential and proprietary data.
"As such, we expect to see businesses actively seeking out these specialised server solutions as an essential component of their strategy to create a secure and high-performance environment for processing and managing their critical data."
Energy
NTT also touched on power consumption in a list of 2024 predictions. "Typical data centre racks consume around six to eight kilowatts of electricity. AI is accelerating that increase in density, and it is now common to see racks anywhere from 50 to 100 kilowatts, and beyond. The expectation is that power consumption will double and triple in the coming years. These racks, which consequently generate more heat, and need more cooling, are playing havoc with enterprises' net zero targets," the company said.
"In 2024, more enterprises will work in close partnership with their energy providers to explore more sustainable options. Tightening data centre regulations will accelerate this disruptive trend, as more governments and subscribers prescribe specific energy efficiency usage and/or supplies, to meet their own sustainability goals."
Storage
"Nearly 90% of data in cloud data centres resides on hard drives. As
data centre infrastructure approach natural refresh cycles, there is an
opportunity to replace fleets of lower-capacity hard drives with higher
capacity drives," noted Seagate in a list of 2024 predictions.
"The
average hard drive capacity in the data centre today is 16 TB based on
conventional PMR technology. The new HAMR technology can physically
store more bits on a single drive; essentially it will enable data
centre managers to populate those same slots with 30 TB high-density
drives, yielding significant power and space savings and massive TCO
efficiencies—including CPU, RAM, and floor space.
"On top of
that, HAMR technology will also help data centre operators to further
reduce their carbon footprint with more durable and energy efficient
storage, playing an even more influential part in tackling the 53.6
million metric tons that make up the global e-waste mountain."
Governance
Maiolo from OVHcloud said: "Countries across Asia are intensifying efforts to develop data governance strategies and legislation, and enforce new laws focused on data privacy, security, and protection. Thailand, Vietnam and Indonesia, for example, have either adopted or are in the process of implementing comprehensive privacy laws.
"Despite that, disparities in data protection policies among countries persist – these variations create a complex regulatory landscape, leaving companies perplexed about differing enforcement practices worldwide and impeding local customers from fully leveraging cloud services. Particularly, in the context of the APAC digital economy where data serves as the lifeblood, ensuring unimpeded and secure data flows through the cloud will be crucial for the region's success.
"Ultimately, these considerations will see data sovereignty grow to become a driver of data storage trends in 2024 and beyond. Given the intricacies involved, Asia's progression in embracing cloud solutions has to emphasise transparency."
"Cloud providers need to offer businesses enhanced support through local compute and storage capabilities to meet the evolving data compliance needs. Transparency must be the established norm sought by businesses, alongside with cloud providers willing to adhere to this standard," Maiolo concluded.
Sustainability
Peluffo said increasing workloads and power demands will put additional pressure on power, carbon and water usage effectiveness, with a renewed focus on PUE, CUE and WUE metrics to measure and manage data centre sustainability.
"To achieve their key performance and sustainability objectives, operators should make sure that their facilities, and in particular their network infrastructure, remain flexible and scalable to meet future operational demands. A well-designed high-performance fibre infrastructure can help deliver improved power efficiency and increased flexibility to help next-gen data centres in future proofing for terabit speeds while improving sustainability performance," he said.
"Today, tools like predictive analytics are already enabling the automation of operations, improved resource management, and enhanced energy and water usage efficiency in line with aggressive sustainability performance targets. AI-based solutions can also deliver insights that improve operational decisions like workload management and capacity planning.
"In 2024, as environmental concerns take centrestage, sustainability is becoming a non-negotiable aspect of data centre development. As such, numerous companies and governments have implemented ambitious sustainability goals, including Singapore," said Kraan.
"The nation’s commitment to energy efficiency, coupled with innovative cooling technologies will likely shape the region in the coming year."
Maiolo also commented on the anticipated environmental impact of AI. "The computational intensity and server computer density required by AI generate significant heat – with estimates that suggest that training certain AI models, such as natural language processing, can produce carbon emissions equivalent to building and driving five cars over their lifetimes. This underscores the pressing need for greener data centres equipped with the necessary power and scalability to address the growing carbon footprint associated with the increasing reliance on emerging technologies," he said.
"After all, as the Asia-Pacific region scales its digital ambitions, it will also have to balance the carbon impact of its data centres. Power, water, and carbon usage are all critical metrics in assessing the efficiency of data centres. Fortunately, water cooling technology has proven itself as a far more energy-efficient and sustainable way of cooling, eliminating the need for air conditioning – with significant benefits to costs and environmental impact. Collaborating with efficient providers, especially in light of the increase in AI workloads, becomes increasingly valuable. Such partnerships can pave the way for a clear, bright trajectory toward a more environmentally-responsible future."
*HAMR is heat-assisted magnetic recording, and PMR is perpendicular magnetic recording. TCO refers to total cost of ownership, and CPU to central processing unit. RAM is an acronym for random access memory.
*PUE, CUE, WUE stand for power usage effectiveness, carbon usage effectiveness and water usage effectiveness respectively.
E is for edge computing
 |
| Source: Milestone Systems. Toft. |
"2024 will see a major acceleration in AI development on the edge, particularly in devices like cameras and sensors. Until now, AI tasks were processed either in the cloud or in a limited way on local devices, but there is now a middle ground with the rise of edge AI.
"The rise of edge AI allows for more independent and efficient processing, reducing reliance on cloud resources. This is especially advantageous for real-time applications in devices like smart cameras and IoT devices that can analyse and respond to data in real-time, providing cost-efficiency by saving bandwidth costs and reducing latency," said Malou Toft, VP Asia Pacific, Milestone Systems.
"Overall, the enhanced capabilities and cost-effectiveness of edge AI position it as a compelling security solution for the future, with the potential for unlocking more sophisticated applications in devices."
"The integration of edge computing with Internet of Things (IoT) technologies is pushing the boundaries of data storage, demanding unprecedented levels of scalability, latency, and operational flexibility to correlate real-time data with vast historical datasets for immediate analysis and forecasting. For instance, digital twins, virtual presentations of physical systems like cities or industrial complexes require both real-time and historical data for accurate simulations and predictions," noted Seagate in its 2024 predictions.
NTT's 2024 predictions include one about the combination of IoT, private 5G, and edge computing enabling real-time insights and better decisions. "With enterprises accelerating digitisation efforts, more connectivity and even more devices are needed as enterprises continue to digitise their physical environments," the company said.
"The edge will significantly grow in importance as enterprises need data to feed analytics platforms powered by AI/machine learning (ML). Increased automation due to labour shortages, computer vision, and digital twins will be key use cases driving the need for robust edge capabilities.
"To go further, organisations need outside help, with eight in 10 expecting their dependency on third-party edge services to grow over the next two years."
 |
| Source: Boomi. Irecki. |
Data insights will be delivered at the point of capture in future, but more needs to be done before that can happen effectively, said David Irecki, Director of Solutions Consulting, APJ, Boomi. "Organisations will need to invest more in metadata (and its management) to avoid having data swamps and the problem of data scientist teams having to scramble to decipher the contents of their data lakes into business values.
"AI will play a major role in this, providing ‘insights’ that the business can ‘action’ upon in more real-time than previously. This is even more imperative as we continue to look towards embracing industry 4.0 technologies for data insights, leveraging automation, data analytics, and artificial intelligence to forge smarter factories."
Securing the edge is also important. "As businesses pursue heightened operational efficiency, the integration of non-human entities such as robotic process automation (RPA), physical robots, and Internet of Things (IoT) systems into the workforce is on the rise – with the APAC region at the frontline of IoT adoption," said Chern-Yue Boey, Senior VP, APAC, SailPoint.
 |
| Source: SailPoint. Boey. |
"Despite the productivity gains, these entities introduce new security challenges, particularly with weak credential controls in many IoT devices, making them even more susceptible targets for cyberattacks. In fact, our recent report shows that more than 30% of identities in an organisation are not properly covered by identity solutions – with gaps around third-party identities, machine identities and data.
"With that, the imperative for businesses will then be embracing the next generation of identity security solutions, fuelled by automation and machine learning."
"Traditional security is akin to building a higher and higher wall to keep out intruders, but once the intruder is inside it’s very difficult to catch them. In 2024 expect a rise of breaches where attackers have exploited an existing vulnerability and are able to move laterally through a network," said Vectra AI in a list of 2024 predictions.
"Security incidents will move away from compromised endpoints, ushering in a new era of threats primarily targeting federated identity systems, public clouds, and business-email-compromise (BEC). This new breed of attacks will exploit the vulnerabilities and relative immaturity of security practices related to cloud, identity and SaaS applications."
*SaaS stands for software as-a-service.
F is for fantasy & fraud
Philip Sellars, Machine Learning Researcher, Darktrace said that AI hallucinations have to be addressed. "As generative AI tools push into widespread business use and move towards active decision making, not just answering questions, hallucinations will become a real issue. ," he said.
"Businesses will have to examine their exposure to the risk created by tools that regularly give them false information and the tolerance for mistakes made by an AI that is taking actions on a user’s behalf will be notably lower than it is when answering their questions.
"Consequently, we’ll see a major focus on techniques to minimise hallucinations in 2024. Approaches such as pairing LLMs with other models, data tables or searchable vector databases, restricting the output of LLMs to checkable formats and other methods will become ubiquitous with the deployment of generative AI tools."
"Given recent progress with generative AI–including audio, images, video, and text–threat actors will have additional tools, capabilities, and approaches to create malicious content. The growing ease of creating persuasive text, manipulated videos, and audio using generative AI is blurring the lines between illusion and reality, making it increasingly challenging for voters to discern authentic information. Stakeholders from across the government, the AI field, and the cybersecurity community at large will need to collaborate to monitor developments in this space."
Explore
Read what industry observers think will happen in 2024 from G-I.
No comments:
Post a Comment