Introduction
Chunghwa Telecom is the largest telecom company in Taiwan, offering various services, including domestic and international fixed communication, mobile communication, broadband, and internet services. In addition to these traditional services, the company provides information and communication technology services to enterprise customers with big data, information security, cloud computing, and IDC capabilities. (https://www.cht.com.tw/en/home/cht)
Chunghwa Telecom operates many virtual machines (VMs) in their data centers. Keeping the resilient operation of the VMs while identifying opportunities to reduce wasteful resources is one of the critical operational challenges faced by the company. This challenge requires Chunghwa Telecom to right-size VM resources accurately, providing a roadmap for cost savings while contributing to the sustainability goal of their infrastructure. However, traditional VM monitoring tools only provide passive resource usage. It is nearly impossible to achieve multiple goals without a solution that provides accurate, application-aware prediction spanning across layers of the infrastructure and resource recommendations to meet application demand in real time.
This case study explores Chunghwa Telecom’s success in achieving cost savings and Green IT objectives by leveraging ProphetStor’s Federator.ai, including the reduction of predicted idle resources in VM clusters and accurate provisioning of VM resources for operation resilience. Additionally, the implementation process and results achieved through using Federator.ai will be discussed.
Solution: Keep VM Resources Right-size and Minimize Waste
Chunghwa Telecom had to ensure that its mission-critical applications in the data centers performed optimally and met its customers’ service-level agreements (SLAs) while also optimizing costs, given the large number of VMs deployed. The company addressed the challenges of costs and operation resilience by adopting ProphetStor’s Federator.ai, an intelligent IT operations (AIOps) platform that understands operation metadata and metrics collected by existing monitoring services, performs workload dynamics predictions, and recommends Just-in-Time Fitted resources for applications.
Federator.ai’s patented multi-layer causality analysis enables the identification of causal correlations between mission-critical applications and the utilized resources, providing crucial insights for optimization. The platform uses advanced AI-powered technology to accurately forecast resource usage and continuously right-size VM resources based on their dynamics.
Federator.ai’s ability to provide continuous intelligent resource orchestration for Kubernetes container resources on VM or bare metal allows Chunghwa Telecom to keep its VM resources properly sized for operational resilience. Additionally, Federator.ai’s machine learning-based resource recommendation engine allows Chunghwa Telecom to provision VM resources more accurately and reduce operational costs while working towards its low-carbon IT infrastructure goals. The multi-layer operation models in Federator.ai can be used for short-term adjustments or long-term planning, providing a roadmap for cost savings and contributing to the infrastructure’s sustainability goals.
Moreover, Federator.ai provided insights into how resources were needed for the next few hours to the next few months, allowing Chunghwa Telecom to have a clear picture of resource usage for different cloud deployments. This made it easy to execute quick resource adjustments or long-term resource planning, enabling Chunghwa Telecom’s operation team to make intelligent actions that keep applications running smoothly and continuously.
Implementation: Seamless Integration into Infrastructure with Great Support
Results: Embracing Green IT and Cost Savings by Reducing Idle Resources
After implementing ProphetStor’s Federator.ai solution, Chunghwa Telecom experienced significant improvements in resource utilization and cost savings, thus further strengthening their commitment to ESG (Environmental, Social, and Governance) as a leading brand with progress towards their Green IT goals. Federator.ai’s dynamic workload predictions accurately forecasted resource usage, allowing for continuous right-sizing of VM resources. Additionally, its machine learning-based resource recommendation engine allocated the right amount of VM resources in time, minimizing idle resources and enhancing resource utilization and ultimately leading to improved operational resilience and cost savings. As a result, the total number of servers required was reduced, resulting in approximately 30% cost savings. These results demonstrate the effectiveness of Federator.ai in helping Chunghwa Telecom optimize their VM resource allocation and achieve their goals of cost savings and sustainability.
Chung-Shuo Lin
Managing Director of Chunghwa Telecom Cloud System Department
Conclusion
30%
Reduce the total number of servers required and save costs by approximately 30%