1 of 39

N2P: A National Research Cloud Service Tailored for Brazilian Scientists

Sergio Leal – RNP (Brazil) sergio.fonseca@rnp.br

Lucas Ma – RNP (Brazil) lucas.ma@rnp.br

Good morning everyone, and thank you very much for the opportunity to be here today. I’m very pleased to share the work we have been developing at RNP in Brazil around N2P. This initiative represents our vision for a national research cloud platform specifically designed to support the needs of Brazilian scientists, universities, and research institutions. Over the last few years, we have been exploring how advanced networking, cloud infrastructure, digital identity, managed services, and more recently AI-driven automation can be combined into a unified ecosystem for research and innovation. The central idea behind N2P is that researchers should not need to become cloud infrastructure specialists in order to access advanced computational capabilities. Instead, the infrastructure should become more accessible, more secure, more intelligent, and more aligned with scientific workflows. Today I will walk through the challenges we identified, the services we delivered, and how we are now evolving toward an agentic and sovereign AI-enabled research cloud model.

2 of 39

Agenda

2

To structure the discussion, I divided the presentation into three main parts. First, I’ll explain what N2P is and why we believed Brazil needed a national research cloud initiative focused specifically on researchers rather than only on infrastructure procurement. In this first section, I’ll also share the research we conducted with scientists and institutions to understand their operational challenges and expectations around cloud adoption. In the second part, I’ll present what we have already delivered, including the service lifecycle, onboarding strategy, consulting model, governance approach, managed support capabilities, and operational tooling. Finally, in the third section, I’ll discuss what we call Agentic N2P, where AI agents become active participants in infrastructure operations, optimization, governance, and automation. This final part is especially important because it reflects how we believe research infrastructure will evolve over the next several years, combining human expertise with intelligent autonomous systems.

3 of 39

-

3

Before talking about the platform itself, it’s important to understand who we are at RNP and the role we play in Brazil’s research ecosystem. RNP is the Brazilian National Research and Education Network, and our mission extends far beyond providing connectivity. We operate as a national digital collaboration platform that connects universities, laboratories, public institutions, researchers, educators, and innovation centers across the country. Our objective is to promote innovation in information technology and democratize access to advanced digital capabilities. We support collaboration by connecting people, institutions, ideas, and services through a trusted infrastructure environment. In practice, this means enabling researchers in very different regions of Brazil to access the same digital opportunities, collaborate internationally, and leverage advanced computational resources regardless of the local limitations of their institutions. This nationwide collaborative mission became one of the main motivations behind the creation of N2P.

4 of 39

-

4

his slide illustrates the scale and complexity of the RNP ecosystem. What you are seeing is not simply a network map, but a national digital infrastructure that supports education, science, and innovation across Brazil. The platform interconnects hundreds of institutions and research centers through high-capacity backbone infrastructure distributed throughout the country. In addition to national connectivity, RNP also supports international scientific collaboration through strategic global partnerships and exchange points. The reason this matters for N2P is because cloud services become much more valuable when they are integrated into a trusted national ecosystem. Researchers do not only need computing power; they also need secure collaboration, data mobility, institutional interoperability, and reliable access to advanced services. The existing RNP infrastructure gave us the opportunity to think about cloud computing not as isolated infrastructure provisioning, but as a coordinated national research platform capable of supporting long-term scientific development and digital sovereignty.

5 of 39

-

5

6 of 39

-

6

N2P reflects a shift in perspective. Traditionally, advanced academic networks focused mainly on connecting institutions. Our vision was to transform that connectivity into direct value for researchers and scientific communities. The platform connects advanced network infrastructure to people and institutions, leverages public cloud environments to deliver scalable and agile services, enables secure connectivity and integrated digital identity, and ultimately drives innovation and digital transformation. What is important here is that N2P is not simply a cloud marketplace or a procurement platform. It is intended to become an integrated research ecosystem where infrastructure, identity, support, governance, and intelligent automation work together to reduce friction for scientific work. We wanted researchers to focus more on science and less on infrastructure complexity.

7 of 39

7

Before defining the architecture of N2P, we conducted a large design thinking exercise focused on understanding the core problems faced by researchers. We engaged more than 2,100 researchers across different institutions and received direct responses from 117 participants. One important characteristic of this audience was the high level of academic maturity. Around 84 percent of respondents had either PhD or postdoctoral backgrounds, which gave us confidence that the feedback represented real advanced research demands. Another extremely important finding was that 88 percent considered computational resources essential to their scientific activities. This confirmed that cloud and advanced computing are no longer optional capabilities in modern research. We also identified strong representation from STEM disciplines, biological sciences, and health sciences, which highlighted the diversity of scientific workloads requiring scalable digital infrastructure. These findings validated the need for a national platform capable of supporting different research domains with flexible and accessible cloud services.

8 of 39

8

One of the most interesting findings from the research was the uneven maturity of cloud adoption among scientists. Approximately 64 percent of respondents were already using some form of cloud infrastructure, which shows that the scientific community clearly recognizes the value of scalable computing. However, 25 percent were still not using cloud services at all, and another 11 percent expressed interest but had not yet started. This gap revealed something very important for us: infrastructure availability alone does not guarantee adoption. Researchers need guidance, onboarding, support, trust, and operational simplicity. Many scientists understand the potential benefits of the cloud but feel uncertain about where to start, how to select services, how to manage costs, or how to maintain security and compliance. This insight strongly influenced our decision to prioritize managed services and support-oriented workflows instead of only providing raw infrastructure access.

9 of 39

9

To better understand the global landscape, we also benchmarked cloud service strategies across forty National Research and Education Networks around the world. What we found was a very rich ecosystem of services already being offered internationally, including IaaS, cloud credits, private peering, HPC integration, training programs, procurement frameworks, advisory services, object storage, and community initiatives. However, we also observed significant fragmentation. Many organizations provided cloud infrastructure institutionally, but relatively few had created integrated researcher-centric experiences that combined access, support, governance, onboarding, and lifecycle management into a unified platform. This benchmarking exercise helped us identify both best practices and gaps in the global ecosystem. N2P emerged from this analysis as an attempt to combine infrastructure delivery with operational guidance, identity federation, governance, and managed support into a more cohesive national platform model.

10 of 39

10

11 of 39

11

12 of 39

12

Financial management emerged as another critical concern. Researchers consistently reported challenges related to cloud cost predictability, budgeting, and financial visibility. Scientific projects often operate under fixed grants and limited funding windows, which means unexpected infrastructure costs can quickly become problematic. Many participants described difficulty forecasting expenses or understanding how consumption patterns translated into operational costs. This created anxiety around cloud adoption because researchers feared losing financial control over their projects. For this reason, FinOps and governance capabilities became strategic priorities within N2P. We wanted to provide researchers with tools and guidance that would make cloud usage more transparent, manageable, and financially sustainable.

13 of 39

13

Security concerns were also extremely prominent throughout our discussions. Researchers and institutions expressed concerns about data protection, compliance obligations, encryption management, and risks associated with public cloud environments. This is especially important in scientific research because projects often involve sensitive data, intellectual property, healthcare information, or strategic national research activities. Researchers wanted the flexibility and scalability of cloud environments, but they also wanted assurance that governance, compliance, and security requirements would be respected. This reinforced our belief that national research cloud initiatives must incorporate strong governance and identity models from the beginning rather than treating them as secondary operational concerns.

14 of 39

14

The final group of pain points involved operational limitations related to funding and technical support. Many institutions lack dedicated cloud specialists, advanced infrastructure teams, or sufficient technical training resources. Researchers often operate under constrained budgets and limited operational staffing, which makes it difficult to deploy and maintain complex cloud environments independently. In practice, this means that even when infrastructure is technically available, adoption may still fail because users do not have enough support to operationalize it effectively. N2P was designed specifically to address these structural limitations through managed services, consulting, onboarding assistance, governance support, and shared operational capabilities that reduce the burden placed on individual research teams.

15 of 39

Global benchmarking with other NRENs

15

After understanding the researchers’ challenges, validating the demand, and benchmarking global practices, the next step was to transform these insights into a practical operational model. This second chapter focuses on what we actually delivered through N2P. Rather than building only a technical platform, we designed a complete service journey that supports researchers from their first contact with cloud services all the way through operational management and eventual service exit. The objective was to reduce complexity, improve accessibility, and create an ecosystem where researchers could focus on scientific outcomes instead of infrastructure administration. In this section, I’ll walk through the lifecycle model, service discovery approach, consulting structure, onboarding process, operational support, governance, and automation capabilities that now compose the N2P environment.

16 of 39

16

The foundation of the operational model is organized around three major phases: access, service discovery, and assessment. We wanted the user experience to feel structured and guided instead of fragmented. The first phase is access, where researchers enter the N2P ecosystem through secure identity federation and integrated authentication services. The second phase is service discovery, where users can explore available infrastructure, managed services, catalogs, and AI-assisted recommendations tailored to their research needs. The final phase is assessment, where researchers interact directly with specialists who help evaluate architectures, workloads, governance requirements, and operational strategies. This structure was intentionally designed to simplify cloud adoption and reduce the cognitive burden placed on scientific teams. Instead of expecting researchers to navigate the cloud ecosystem alone, N2P provides a guided and supported experience from beginning to end.

17 of 39

17

In addition to the onboarding journey, we also designed a complete service lifecycle model composed of three operational stages: service acquired, service used, and service terminated. We considered this lifecycle essential because research environments are dynamic. Scientific projects evolve over time, grants change, workloads grow, collaborations expand, and eventually some environments need to be migrated, archived, or decommissioned. Many cloud strategies focus only on provisioning infrastructure, but we wanted to support the full operational journey. During acquisition, researchers receive guidance and onboarding support. During usage, they gain access to operational assistance, financial governance, and optimization services. Finally, during termination, we provide support for migration, workload closure, and controlled service exit. This lifecycle perspective helps institutions maintain continuity, governance, and sustainability throughout the entire duration of research projects.

18 of 39

18

Access is one of the most important elements of the platform because it defines how researchers enter and interact with the ecosystem. We built this layer around three core capabilities. The first is the N2P environment itself, which provides a secure collaborative infrastructure where institutions and researchers can connect and operate. The second is CAFe federated authentication, which allows users to access services seamlessly using trusted institutional identities. This significantly reduces friction because researchers do not need to manage multiple isolated credentials across different systems. The third capability is RNP+, which creates unified identity integration across the research ecosystem. Together, these components establish a secure, interoperable, and scalable access layer capable of supporting collaborative scientific environments across multiple institutions and regions.

19 of 39

19

Once researchers enter the platform, the next challenge is helping them discover and select the right services. This is where the service discovery layer becomes extremely important. We included AI-assisted recommendation mechanisms to help researchers identify infrastructure and platform services aligned with their needs. We also created a centralized service catalog that allows users to explore available capabilities more efficiently. Infrastructure and platform services can be accessed on demand, while managed services provide operational acceleration for teams that do not want to handle low-level infrastructure management directly. Interestingly, some traditional cloud components, such as generic cloud calculators or unmanaged common stacks, were intentionally deprioritized because our research showed that scientists preferred guided and simplified operational experiences. This reflects a broader philosophy behind N2P: reducing operational complexity instead of transferring it directly to researchers

20 of 39

20

Assessment is where human expertise becomes integrated into the platform experience. One of the strongest lessons from our research was that researchers value trusted technical guidance when making infrastructure decisions. Through RNP+ support and technical consulting services, researchers can work directly with specialists who understand both cloud operations and scientific workloads. These specialists help institutions evaluate architectures, estimate resource requirements, understand governance implications, optimize operational models, and accelerate adoption strategies. This assessment layer creates confidence and reduces uncertainty during cloud adoption. Instead of researchers making isolated decisions without support, they can collaborate with infrastructure experts who help align technology choices with scientific and institutional objectives.

21 of 39

21

The acquisition phase focuses heavily on onboarding and operational preparation. When a researcher or institution decides to adopt a service, the process is not treated as a simple infrastructure transaction. Instead, we provide consulting support, onboarding assistance, and managed operational guidance. Researchers can work directly with specialists who help configure environments, define workflows, validate requirements, and accelerate deployment. The idea is to reduce the time between infrastructure acquisition and productive scientific usage. Many research teams do not have the resources or expertise required to independently deploy complex cloud architectures. By offering onboarding and managed support, N2P helps researchers move more quickly from procurement into active scientific work.

22 of 39

22

Once services are operational, attention shifts toward governance, optimization, and sustainability. Two major capabilities become central during this phase. The first is FinOps, which helps researchers monitor and optimize infrastructure spending. Financial governance is particularly important in academic environments because research projects often depend on fixed grants and limited funding cycles. The second capability is specialized cloud support, where operational experts help research teams optimize workloads, improve scalability, solve technical issues, and maintain infrastructure performance. This operational support model reduces the burden on scientists and allows them to focus more directly on research outcomes rather than day-to-day infrastructure management. The broader goal is to create a sustainable operational environment where cloud adoption remains predictable, efficient, and aligned with institutional priorities.

23 of 39

23

Eventually, some projects conclude, workloads change, or institutions decide to migrate services. That is why the termination phase is also treated as an essential part of the lifecycle. Service exit is often neglected in many cloud strategies, but it becomes critically important for governance, compliance, data preservation, and operational continuity. Researchers may need to archive datasets, migrate workloads, preserve reproducibility, or decommission infrastructure in controlled ways. Through N2P, we provide support for this transition process so institutions can maintain visibility and governance even during service closure. Treating termination as a managed operational phase helps avoid fragmentation, reduces operational risk, and supports long-term sustainability across the research ecosystem

24 of 39

Performance Datadog

24

To increase extensibility and operational flexibility, we also designed a modular tooling ecosystem based on a CLI, plugins, and an application marketplace. The concept here is very important because scientific environments evolve continuously. Researchers may need analytics tools, deployment automation, security modules, database utilities, or workflow integrations depending on their specific domains and projects. By creating a plugin-based architecture, we allow the platform to adapt over time without forcing users into rigid operational models. The CLI enables automation and scripting, while the application marketplace simplifies installation and integration of capabilities directly into research workflows. This architecture also reinforces institutional control because organizations can manage which tools are available and how they are operationalized within the ecosystem.

25 of 39

Performance Datadog

25

Behind the user-facing experience, orchestration becomes a critical technical component. In this architecture, we use n8n as an orchestration layer connecting web interfaces to CLI components and infrastructure automation workflows. Through secure APIs, the platform integrates technologies such as Git, Docker, Kubernetes, and AWS CLI into a coordinated operational environment. What makes this important is that researchers can interact through simpler interfaces while still benefiting from sophisticated backend automation capabilities. The orchestration layer acts as a bridge between usability and infrastructure complexity. It allows automation pipelines, deployment workflows, governance controls, and operational tasks to be coordinated in ways that remain accessible to scientific users without requiring them to become infrastructure engineers themselves.

26 of 39

-

Agentic N2P

26

Up to this point, the discussion has focused on the current N2P platform and the services we have already delivered. But the next phase of evolution is even more transformative. We are now moving toward what we call Agentic N2P, where AI agents become active participants in infrastructure operations, optimization, governance, and decision support. The reason this matters is because modern cloud environments are becoming too dynamic and complex for purely manual operations. Researchers increasingly need intelligent systems capable of monitoring infrastructure, detecting anomalies, optimizing workloads, enforcing governance, and automating repetitive operational tasks. In this final section, I’ll explore how we are approaching autonomous agents, governance models, and AI-driven operational infrastructure in a secure and sovereign way.

27 of 39

What’s sovereignty?