Running a Pre-exascale, Geographically Distributed, Multi-cloud Scientific Simulation

<title>Abstract</title> Serverless computing reduces management overhead and scales automatically, yet its latency and performance differ significantly across cloud platforms. This study evaluates serverless function execution and infrastructure-level characteristics on Amazon Web Services (AWS), Microsoft Azure, and Google Cloud Platform (GCP). We benchmark function invocation latency in edge deployments, conduct regional latency comparisons across AWS and Azure, and use PerfKit Benchmarker to measure network throughput, network latency, and storage I/O on AWS and GCP. Results show that Azure recorded the lowest minimum latency (555 ms) but suffered from high tail delays, while GCP delivered the most consistent average performance (1.14 s). AWS exhibited moderate latency with a relatively stable distribution. At the infrastructure level, GCP nearly doubled single-stream throughput (9.35 Gb/s) compared to AWS (4.97 Gb/s), whereas AWS achieved slightly lower round-trip network latency (60 vs. 67 $\mu$s). Storage I/O performance was nearly identical across both providers. These findings link infrastructure characteristics to observed serverless behavior, providing actionable insights for latency-sensitive and multi-cloud application deployments.

Serverless computing has fundamentally transformed cloud technology by enabling developers to deploy applications without the burden of managing server infrastructure. This paper delves into the serverless offerings from major cloud providers—AWS (Amazon Web Services), Azure (Microsoft Azure), and GCP (Google Cloud Platform)—and explores how each contributes to this paradigm shift. The discussion highlights the core advantages of serverless computing, including cost efficiency through a pay-as-you-go model, automatic scaling that adjusts to changing demands, and reduced operational complexity, which allows developers to focus more on coding and less on infrastructure concerns. However, serverless computing is not without its challenges. Notable issues include "cold start" delays, where initial function executions may experience latency, impacting overall performance. Additionally, the shared nature of serverless environments raises potential security concerns, and the limited visibility into the execution environment can make debugging and optimization more complex. The paper also considers future trends in serverless computing across AWS, Azure, and GCP. It discusses emerging approaches such as combining serverless with traditional computing models to create hybrid architectures that leverage the strengths of both. Furthermore, it examines ongoing advancements aimed at improving support for complex, stateful applications, seeking to address current limitations and enhance the capabilities of serverless platforms. This exploration provides a detailed look at the current state of serverless computing and anticipates future developments, focusing on how AWS, Azure, and GCP are shaping the evolution of this technology

The article investigates the impact of modern cloud technologies on the efficiency of DevOps processes, which are key to the automation, flexibility and reliability of software development. With the rapid development of information technology, cloud platforms play an important role in accelerating software product development, testing and deployment. Implementing DevOps in the cloud environment reduces the time to market, increases the stability of software systems and optimizes the use of computing resources, which is critical for modern companies. This paper discusses DevOps teams' main challenges, including the need to scale infrastructure rapidly, ensure continuous integration and delivery (CI/CD), automate testing, monitor performance, and optimize costs. Particular attention is paid to security problems, configuration management and the human factor minimization when implementing program code changes. A comparative analysis of the three leading cloud platforms - Amazon Web Services, Google Cloud Platform, and Microsoft Azure – is carried out in the context of their impact on DevOps processes. The possibilities of automation, support for container technologies, infrastructure scalability, monitoring tools, and integration with CI/CD pipelines are evaluated. AWS was found to offer the broadest set of DevOps tools with a high level of automation, making it an attractive choice for organizations focused on complex enterprise solutions. Google Cloud Platform demonstrates the best support for Kubernetes and containerization, an essential factor for teams working with microservices architecture. Microsoft Azure provides the most profound integration with the Microsoft ecosystem and is the best choice for companies that use Windows and related products. An experimental study has shown that the choice of a cloud platform significantly impacts the speed of release cycles, the stability of the deployed infrastructure, the ability to respond quickly to changes in load, and the productivity of DevOps teams in general. Recommendations for choosing the optimal cloud platform are proposed depending on the specifics of the project, the scale of the organization, the level of system load, and automation requirements.

As enterprises increasingly migrate to cloud platforms, managing the cost-efficiency of data warehouses in environments such as Amazon Web Services (AWS), Google Cloud Platform (GCP), and Microsoft Azure has become a critical concern. This paper proposes a conceptual model for optimizing the financial and operational management of cloud-based data warehouses. Through a synthesis of recent peer-reviewed studies, whitepapers, and real-world implementation reports from 2015 to 2024, the model integrates strategic design principles, workload optimization techniques, and governance frameworks across multi-cloud ecosystems. The proposed model emphasizes dynamic workload management, tiered storage optimization, intelligent scaling policies, and metadata-driven governance to ensure cost control without compromising performance. Key architectural components include serverless and autoscaling compute layers, storage lifecycle management, query optimization strategies, and automated performance tuning mechanisms. Particular focus is placed on the unique features and pricing models of AWS Redshift, GCP BigQuery, and Azure Synapse Analytics, detailing how organizations can exploit platform-specific capabilities to enhance cost-efficiency. Furthermore, the model incorporates modern innovations such as FinOps practices, usage-based cost allocation, predictive scaling powered by machine learning, and real-time cost observability dashboards. It also outlines potential pitfalls, such as overprovisioning, inefficient data partitioning, and underutilized reserved instances, and provides mitigation strategies to address them. By aligning technical architecture decisions with proactive financial operations, this conceptual model offers a pathway for organizations to balance performance, scalability, and budget constraints effectively. The study concludes by recommending future directions, including AI-driven autonomous warehouse management, unified billing optimization across multi-cloud deployments, and frameworks for continuous cost-performance evaluation. Mastering cost-efficient warehouse management is increasingly essential for organizations seeking to maximize the value of their data assets while maintaining fiscal responsibility in complex, distributed cloud environments.

Mainframe systems have a track record of reliability, but they need help in today’s fast-changing business world. These include high upkeep expenses and limited room for growth. This paper looks at three top cloud platforms—Amazon Web Services (AWS), Microsoft Azure, and Google Cloud Platform (GCP)—to update old mainframe programs. We compare them in detail, looking at what each does best. AWS has a wide range of tools to move data. Azure works well with cloud and on-site systems and fits into company setups. GCP leads in AI and machine learning. Azure is the top option for updating mainframes, offering compatibility with various cloud configurations, integration with current company infrastructure, and sophisticated data analysis driven by AI. These conclusions provide valuable information for companies transitioning mainframe systems to the cloud, enabling expansion, reduced expenses, and future-ready IT systems.

Scientific high throughput computing needs are growing dramatically with time and public Clouds have become an attractive option for occasional bursts, due to their ability to be provisioned with minimal advance notice. The available capacity of both compute and networking is however not well understood. This article presents the results of several production runs of the IceCube collaboration that temporarily expanded its batch system environment with GPU-providing compute instances from the three major Cloud providers, namely Amazon Web Services, Microsoft Azure, and the Google Cloud Platform. The aim of these Cloud bursts was to push the limits of Cloud compute, with a particular emphasis on GPU-providing instances. On the compute side, we showed that it is possible to reach peaks of over 380 fp32 PFLOPS using all available GPU-providing instance types and integrate over 1 fp32 EFLOP hour in a single workday by using only the most cost-effective ones. On the network side, we showed intra-Cloud network throughputs of over 1 Tbps, and 100 Gbps throughputs toward on-prem storage both using shared peering arrangements and dedicated network links.

The explosion of data in the digital era has posed major challenges handling, computing and analyzing enormous and complex datasets. Cloud computing has arisen as a revolutionary solution providing scalable and elastic infrastructure necessary to deal with incoming big data workloads. This study employs an empirical approach to evaluate the performance, cost efficiency, and scalability of the three dominant cloud service models, Infrastructure-as-a-Service (IaaS), Platform-as-a-Service (PaaS) and Function-as-a-Service (FaaS) on Amazon Web Services, Microsoft Azure and Google Cloud Platform. Standard big data analytics workloads running real-time stream processing and machine learning activities were then implemented in Apache Spark, Hadoop, and Kafka on harmonized cloud environments. Key performance metrics such as execution time, CPU utilization, memory, cost per task and throughput were taken, analyzed statistically using ANOVA and Tukey’s post hoc tests. Results show that FaaS configurations are always faster in execution speed, memory efficiency and cost compared to IaaS, while IaaS delivers better CPU usage for continual workloads. AWS and GCP platform performed relatively balanced when compared to Azure. It is concluded that serverless architecture is, in fact, optimal for modular and burst-oriented analytics, and hybrid models might be more appropriate for complex pipelines. These results can offer cloud architects practical directions towards scalable and cost-effective big data solutions.

SummaryCloud‐based services have proved useful in several research fields, such as engineering, health science, and astrophysics, to mention a few examples. The computational environmental science community developed a strong need for cloud facilities to store, process, and manage data from observations and numerical models for simulations and forecasts. Weather forecast models and global sensor networks deal with multidimensional geo‐referenced data∖sets. However, environmental data consumer applications usually require a relatively small amount of multidimensional input data slice to analyze a specific area or time interval. Hence, reducing data dimension for information retrieval is mandatory. This paper presents a twofold solution: a technique to load and retrieve the sliced multidimensional data set on different cloud services such as Amazon Web Service (AWS), Google Cloud Platform, and Microsoft Azure. The experimental results performed on these cloud services highlight that the proposed method can significantly speed up the process of loading and retrieving the data slices compared to working with the entire data set in bulk or OPeNDAP server.

Cloud computing has caused a revolution in how companies do business giving them flexibility, scalability, and cost-efficiency like never before. It lets organizations use computing resources through the internet instead of relying on the usual in-house setup making cloud computing a key part of today's IT plans. From new startups to big global companies, businesses are using cloud solutions more and more to boost their ability to adapt cut down on running costs, and come up with new ideas. The cloud ecosystem is big and varied, with different deployment models and service types to suit various business needs. Public, private, and hybrid clouds are the three main deployment models each with its own pros and cons. public clouds, like those from Amazon Web Services (AWS), Microsoft Azure, and Google Cloud Platform, offer affordable scalability and global reach. Private clouds, on the other hand, are for organizations that have strict security or compliance needs giving them dedicated environments. Hybrid clouds mix the best parts of both letting businesses find a balance between security and scalability by combining private and public cloud resources

Cloud computing has revolutionized analysis of large datasets. This tutorial provides a comprehensive, practical guide for research groups seeking to leverage cloud platforms for data analysis. The tutorial covers the foundations of cloud computing, including its history, rationale, and use cases for research, followed by detailed comparisons of the 3 major platforms: Google Cloud Platform (GCP), Amazon Web Services (AWS), and Microsoft Azure. Side-by-side comparisons of services, costs, ease of use, and selection guidelines assist researchers in choosing the most appropriate platform. A complete step-by-step example using hospital price transparency data demonstrates the entire workflow from account creation through results retrieval, enabling researchers to begin productive cloud-based analysis within hours.

Rapid expansion in cloud event technologies such as Amazon Web Service's Lambda, IBM Bluemix's OpenWhisk, Google Cloud Platform's Cloud Functions, and Microsoft Azure's Functions motivates study of software development in these services and their potential as a disruptive force in commercial cloud technologies. In addition to discussing the current state of cloud event services, this paper presents two real world applications utilizing these platforms: Lambdefy, a library designed to make traditional web application run effectively in AWS Lambda, and a performant media management service designed by Trek10, capable of resizing thousands of images per second. Furthermore, we discuss how cloud event technologies enable and/or limit these applications, motivate new software design paradigms in a cloud event environment, and highlight compelling use case scenarios and barriers to entry for cloud event services. AWS cloud technologies are exclusively used due to their maturity and the recent release of the other platforms, while Node.js and the Serverless Framework are utilized for deployment and application development.

Modern Software-as-a-Service (SaaS) platforms require a strongand reliable infrastructure that can function seamlessly across various cloud providers, without losing consistency and efficiency in operations. The multi-cloud architecture, however, brings in a lot of complexity due to the differences in provider APIs, resource models, and operational characteristics. When infrastructure management is handled manually, configuration drift becomes a significant issue. The conventional deployment methods are not efficient enough to work at a large scale across different cloud environments. The piece lays out an intricate automation framework that combines Infrastructure as Code (IaC) provisioning, multi-level validation testing, and centralized observability in a flawless manner. Using Terraform, one can carry out declarative infrastructure specifications that can be spread over Amazon Web Services (AWS), Microsoft Azure, and Google Cloud Platform (GCP). Puppet ensures that the standard configurations are kept in the various node populations. Unit testing is employed to confirm the correctness of the individual modules before their integration. Acceptance testing simulates the end-to-end workflows in temporary environments that are exact replicas of production topologies. Deployment of catalog comparison strategies uncovers unintentional configuration changes prior to implementation. Centralized logging collects the events from distributed systems that are open for cross-cutting analysis. Monitoring dashboards draw the time-series metrics from the infrastructure components and the applications. Automated alerting pinpoints the operational issues facing the threshold conditions and anomaly detection. The combination of the framework fills in the critical gaps of the existing single-cloud automation approaches. Declarative specifications are there to remove the complexities of provider-specific. Automated validation is there to stop the propagation of faulty configurations across distributed systems. Unified observability is the consistency of visibility across fragmented multi-cloud landscapes. The outline exhibits the solutions for the realization of enterprise-scale cloud operations by the integration of strategic automation.

This research aims at examining the strategies of performance tuning in cloud computing with emphasis on the optimization of applications, minimized response time, and optimal, affordable resource utilization. The research therefore includes a systematic literature review together with quantitative findings through empirical testing of the proposed model on Amazon Web Services (AWS), Microsoft Azure, and Google Cloud Platform (GCP), in addition to qualitative insights of experts. Auto-scaling, load balancing caching, database optimizing, integration with edge computing and predictive workloads using Artificial Intelligence are the aspects that are also studied as key performance tuning latter. Soon, the investigation, which was made based on the results from applying the four techniques on different applications and two clouds, demonstrates the strength of each technique in achieving different goals. While auto-scale and load balance feature is very helpful in control of workload fluctuations, the caching and database optimization helps in the efficient retrieval of the data. Edge computing reduces latency in response to real-time applications, and the application of artificial intelligence in workload forecast smoothes resource utilization in environments with a rapidly changing workload. Accordingly, the research has shown need for careful choosing of suitable performance-oriented interventions to enhance the application’s interactions, decrease CPU utilization, and cut costs in a cloud environment. Lastly, this work offers practical knowledge about the methods of cloud performance tuning to support better application deployment in the cloud environments.

Cloud Service Providers, exemplified by industry leaders like Google Cloud Platform, Microsoft Azure, and Amazon Web Services, deliver a dynamic array of cloud services in an ever-evolving landscape. This sector is witnessing substantial growth, with enterprises such as Netflix and PayPal heavily relying on cloud infrastructure for various needs such as data storage, computational resources, and various other services. The adoption of cloud solutions by businesses not only facilitates cost reduction but also fosters flexibility and supports scalability. Despite the undeniable advantages, concerns surrounding security and privacy persist in the realm of Cloud Computing. Given that Cloud services are accessible via the internet, there is a potential vulnerability to unauthorized access by hackers or malicious entities from anywhere in the world. A crucial aspect of addressing this challenge is the implementation of robust security measures, particularly focusing on data protection. To safeguard data in the Cloud, a fundamental recommendation is the encryption of data prior to uploading. Encryption should be maintained consistently, both during storage and in transit. While encryption enhances security, it introduces a potential challenge for data owners who may need to perform various operations on their encrypted data, such as accessing, modifying, updating, deleting, reading, searching, or sharing them with others. One viable solution to balance the need for data security and operational functionality is the adoption of Searchable Encryption (SE). SE operates on encrypted data, allowing authorized users to perform certain operations without compromising the security of sensitive information. The effectiveness of SE has notably advanced since its inception, and ongoing research endeavors aim to further enhance its capabilities. This paper provides a comprehensive review of the functionality of Searchable Encryption, with a primary focus on its applications in Cloud services during the period spanning 2019 to 2023. Additionally, the study evaluates one of its prominent schemes, namely Fully Homomorphic Encryption (FHE). The analysis indicates an overall positive trajectory in SE research, showcasing increased efficiency as multiple functionalities are aggregated and rigorously tested.

All traditional sectors are being transformed to a large extent as technology advances. Some examples are smart homes, smart systems, smart playgrounds, smart cities, and so on. There is no exception in the education sector. The modern smart classroom concept is not new, but it is much more effective and efficient using cloud based systems. Machine learning, virtual reality, and augmented reality serve as foundations for creating a modern classroom environment. The data created by these technologies necessitates a large amount of storage space. The cloud makes it possible to do so. The need for latency can be met by choosing the closest datacenter. Users will be able to collaborate closely with resources in the modern environment. Users will be able to work closely with multiple resources in the current environment, and they will have access to them whenever they want. Amazon Web Services, Microsoft Azure, Google Cloud Platform, and others are examples of cloud service companies. Cloud service providers ensure that data is available to users at all times and in all places.