Data analytics, AI, and ML play crucial roles in optimising data centre performance and efficiency. Here’s how these technologies contribute
Data centres have emerged as a crucial component of the IT infrastructure of businesses. They handle vast amounts of data generated by various sources, and over the years have transformed into massive and complex entities. Of late, data analytics has emerged as a necessary ally for data centre service providers, powered by the growing need to improve parameters like operational efficiency, performance, and sustainability. In this blog, we will discuss the different ways in which data analytics and AI/ML can help enhance data centre management and empower data centre service providers to deliver better service assurance to end customers.
How data analytics and AI/ML can help service providers in data centre optimisation
Today, data centre service providers are leveraging data analytics in various ways to optimise data centre operations, reduce costs, enhance performance, reliability and sustainability, and improve service quality for customers. They employ a variety of methods to collect data from colocation, on-premise and edge data centres, which include physical RFID/EFC sensors, server, network and storage monitoring tools, security information and event management (SIEM) systems, configuration management databases (CMDBs), API integration, and customer usage data. The data collected is then fed into a centralised monitoring and analytics platform, which uses visualisation tools, dashboards, and alert systems to analyse the data and generate insights.
Furthermore, by integrating IoT and AI/ML into data centre operations, service providers are gaining deeper insights, automating various processes, and making faster business decisions. One of the most critical requirements today is for analytical tools that can help with predictive assessment and accurate decision-making for desired outcomes. This is achieved by diving deep into factors such as equipment performance, load demand curve, and overall system performance, as well as intelligent risk assessment and business continuity planning. Selection of the right tools, firmware, and application layer plays a major role in making such an AI/ML platform successful.
The relationship between analytics and automation from the perspective of data centres is rather symbiotic. Data centres are already automating routine tasks such as data cleaning, data transformation, and data integration, helping data centre service providers free up resources for more strategic analytics work, such as predictive modelling, forecasting, and scenario planning. In turn, data analytics provides valuable insights that enable data centres to implement intelligent automation and optimisation techniques. This may include workload balancing, dynamic resource allocation, and automated incident response.
Here are some of the key areas where data analytics and automation have a significant impact:
• Enhancing operational reliability: Data analytics, AI/ML and automation can enable data centres to ensure optimal performance. This involves using predictive maintenance, studying equipment lifecycles for maintenance, and incident history analysis to learn from past experiences. In addition, AI/ML-driven vendor performance evaluation and SLA management incorporating MTTR and MTBF further strengthen operations. Leveraging these metrics within the ITIL framework helps data centres gain valuable operational insights and maintain the highest levels of uptime.
• Performance efficiency: Data centres consume a substantial amount of energy to power and maintain desirable operating conditions. To optimise services, track hotspots, prevent hardware failure, and improve overall performance, modern data centres analyse data points such as power usage, temperature, humidity, and airflow related to servers, storage devices, networking equipment, and cooling systems. Prescriptive analytics can take this a step further by providing recommendations to optimise utilisation and performance.
• Predictive maintenance: Predictive analytics is a powerful technology that uses data to forecast future performance, identify and analyse risks and mitigate potential issues. By analysing sensor data and historical trends, data centre service providers can anticipate potential hardware failures and perform maintenance before they escalate, with advanced predictive analytics enabling them to improve equipment uptime by up to 20%.
• Capacity planning: Businesses today must be flexible enough to accommodate capacity changes within a matter of hours. Data centre service providers also need to understand current usage metrics to plan for future equipment purchases and cater to on-demand requirements. Data analytics helps in optimising the allocation of resources like storage, compute, and networking while meeting fluctuations in customer needs and improving agility.
• Security and network optimisation: Data centres can use analytics to monitor security events and detect vulnerabilities early to enhance their security posture. By analysing network traffic patterns, data analytics tools help identify unusual activities that may indicate a security threat. They can also monitor network performance, identify bottlenecks, and optimise data routing.
• Customer insights: Datacentres collect usage data, such as the number of users, peak usage times, and resource consumption, to better understand customer needs and optimise services accordingly. Analytics helps providers gain insights into customer behaviour and needs, enabling them to build targeted solutions that offer better performance and value. For example, through customer-facing report generation, organisations and end-customers can gain valuable insights and optimise their operations. Additionally, analytics accelerates the go-to-market process by providing real-time data visibility, empowering businesses to make informed decisions quickly and stay ahead of the competition.
• Environment sustainability and energy efficiency: Data centres have traditionally consumed significant power, with standalone facilities consuming between 10-25 MW per building capacity. However, modern data centre IT parks now boast capacities ranging from 200-400+ MW. This exponential growth has led to adverse environmental impacts, such as increased carbon footprint, depletion of natural resources, and soil erosion. Using AI/ML, performance indicators like CUE (Carbon Utilisation Effectiveness), WUE (Water Utilisation Effectiveness), and PUE (Power Utilisation Effectiveness) are analysed to assess efficiency and design green strategies, such as adopting renewable energy, implementing zero water discharge plants, achieving carbon neutrality, and using refrigerants with low GHG coefficients. For example, AI/ML modelling can help data centres achieve 8-10% saving on PUE below design PUE – helping to balance environmental impact with an efficiency better than what was originally planned.
AI/ML modelling can help data centres achieve 8-10% saving on PUE below design PUE – helping to balance environmental impact with an efficiency better than what was originally planned.
• Asset and vendor performance management: The foundation of the AI/ML platform lies in the CMDB, which comprises crucial data, including asset information, parent-child relationships, equipment performance records, maintenance history, lifecycle analysis, performance curves, and end-of-life tracking. These assets are often maintained by OEMs or vendors to ensure reliability and uptime. AI/ML aids in developing availability models that factor in SLA and KPI management. It can provide unmatched visibility into equipment corrections, necessary improvements, and vendor performance. It can also help enhance project models for expansion build-outs and greenfield designs, accurately estimating the cost of POD (point of delivery) design, project construction, and delivery.
• Ordering billing and invoicing: AI/ML plays a vital role in enhancing the efficiency and effectiveness of order, billing, and invoicing processes. Its impact spans various stages, starting from responding to RFPs to reserving space and power, managing capacity, providing early access to ready-for-service solutions, facilitating customer onboarding, and overseeing the entire customer lifecycle. This includes routine processes such as invoicing, revenue collection, order renewal, customer Right of First Refusal (ROFR) management, and exploring expansion options both within and outside the current facility.
Selecting the right data analytics solution
The implementation of data analytics and automation through AI/ML requires careful consideration as several parameters, such as data quality and level of expertise play a crucial role in delivering efficient end results. To succeed, businesses need to choose user-friendly and intelligent solutions that can integrate well with existing solutions, handle large volumes of data, and evolve as needed.
At Sify – India’s pioneering data centre service provider for over 23 years, we continuously innovate, invest in, and integrate new-age technologies like AI/ML in operations to deliver significant and desired outcomes to customers. We are infusing automation led by AI/ML in our state-of-the-art intelligent data centres across India to deliver superior customer experiences, increased efficiency, and informed decision-making, resulting in more self-sustaining and competitive ecosystems. For example, leveraging our AI/ML capabilities has been proven to lead to over 20% improvement in project delivery turnaround time. Our digital data centre infrastructure services offer real-time visibility, measurability, predictability, and service support to ensure that our customers experience zero downtime and reduced Capex/Opex.
How do Sify’s AI-enabled data centres impact your business?
• Person-hour savings: Automation of customer billing data and escalations resulting in up to 300 person-hour savings in a month.
• Reduction in failures: Predictive approach for maintenance and daily checks yielding up to 20% reduced MTBF, 10% improved MTTR, and 10% reduction in unplanned/possible downtime.
• Cost savings: Improved power/rack space efficiency and savings on penalties to deliver up to an 8% reduction in customer penalties by maintaining SLAs and a 10% reduction in operating cost.
• Compliance adherence: Meeting global standards and ensuring operational excellence and business continuity.
By Girish Dhavale, CTO, Data Centre Services, Sify Technologies Ltd.