Launched in March 2021, the Tri-Agency Research Data Management Policy includes a requirement that Grant recipients […] deposit into a digital repository all digital research data, metadata and code that directly support the research conclusions in journal publications and pre-prints that arise from agency-supported research. The agencies committed to phase-in the data deposit requirement after reviewing institutional RDM strategies and in line with the readiness of the Canadian research community. Following extensive engagement with partners and interest-holders, the agencies released a What We Heard report in July 2025. A cross-sectional study commissioned by the agencies mapping over 200 institutional RDM strategies also provided further insight into national readiness for data deposit. Overall, members of the research community expressed broad support for making explicit the principle that publicly funded research data should be "as open as possible, as closed as necessary"; however, participants also emphasized the need to clarify several aspects of the data deposit requirement to facilitate implementation. This talk will present the proposed implementation approach for the data deposit requirement and, as part of the agencies' consultation on the approach, will direct attendees to an online form to provide written feedback. The agencies plan to announce the finalized implementation approach by the end of 2026. Feedback from DRI Connect attendees is particularly relevant given the importance of RDM services at institutions and at the Alliance in supporting the successful implementation of the requirement. Dominique Roche Senior Policy Advisor and Chair, Tri-Agency RDM Working Group Social Sciences and Humanities Research Council of Canada dominique.roche@sshrc-crsh.gc.ca

After several years of proposals, planning, purchasing, and finally building and deploying, the renewed Arbutus Cloud has gone live to researchers! Come and hear what UVic RCS and ArbutusNT have learned from this massive project, see the exciting new features and improvements that have come with the renewal, and consider how this highly-cost effective and fully sovereign Canadian research cloud informs directions for DRI more broadly in this AI-driven period of change and opportunity.

Research software is foundational to digital research infrastructure (DRI) in Canada, enabling discovery, reproducibility, and innovation across all disciplines. Recognizing research software as a first-class research output, the Digital Research Alliance of Canada has articulated a national strategy focused on strengthening capability, community, and coordination across the research software ecosystem. This panel will explore how recent and planned initiatives are translating that strategy into concrete, sustainable outcomes for Canadian researchers. The session will highlight the Alliance's Research Software AI Enhancement funding initiative, through which over $2.6M was invested in Research Platforms and Portals (RPPs) to integrate artificial intelligence techniques and deliver AI-enabled services to research communities. RPPs play a critical role in lowering barriers to advanced computing by providing discipline-specific interfaces that integrate data, software, and computational resources. Enhancing these platforms with AI capabilities strengthens their long-term sustainability and ensures continued access to state-of-the-art digital research tools. Building on this momentum, the panel will discuss the planned Research Software Maintenance Fund (RSMF), addressing a well-documented gap in sustained support for critical research software identified across the Canadian research community. The discussion will also outline forthcoming research software services, including a specialized AI-focused consulting service, and the vision for a national research software platform providing core development services, domain-specific software, and reusable research software frameworks for workflows, interactive computational environments, simulation toolkits and other functions. Together, these initiatives illustrate a coordinated, national approach to advancing research software capability as a cornerstone of Canada's future DRI ecosystem.

Data spaces are being introduced globally to address challenges of data access and trust. They provide an abstract layer where data providers and users can securely interact in a trusted environment to negotiate data access via machine actionable data policies. The International Data Spaces Association (IDSA) is a non-profit, consensus-based, member-driven organization focused on establishing and promoting standards for data spaces. Since its inception in 2016, IDSA has grown to include almost 200 members from over 30 countries. IDSA's core contributions are the Dataspace Protocol, the IDSA Rulebook and the IDS Reference Architecture Model. The Rulebook provides the basic principles for trusted data sharing through data spaces, including defining their rules, governance mechanisms, and legal basis, which are then implemented through the IDS Architecture defined in the Reference Architecture Model. The IDSA Data Space Protocol is a formal ISO/IEC technical standard that puts into practise the Reference Architecture Model and Rulebook. Together, these resources support the creation of interoperable and scalable data spaces that enable trustworthy data sharing across industries and borders. The IDSA materials have informed the purpose and positioning of the National Data Spaces Program and will inform the definition of a Canadian National Data Space. The Alliance is a member of the IDSA and engages with their working groups. This workshop will briefly introduce the IDSA materials and how they map onto research community and DRI needs. The main component of the workshop will be to investigate what components of the IDSA materials require adaptation for the Canadian context and what those adaptions should include. The format will include open and group discussions with interactive online tools. Prior knowledge of the IDSA materials is not required, and participants working across DRI organizations are encouraged to attend to ensure a range of perspectives.

"Despite the growing number of formal digital humanities degrees (Sula, Hackney, and Cunningham 2017; Walsh et al. 2021), the … majority of digital humanities training has long occurred outside traditional curricula as complementary initiatives." Canada has long been a leader in digital humanities (DH), yet until a couple of years ago, workshops across the country attracted regional, national, and international participants but lacked formal coordination and certification. Before 2023, there was limited coordination across disparate DH courses, and no official certification process to acknowledge the time and labour of those pursuing DH training. Thus, the Canadian Certificate in Digital Humanities (cc:DH/HN) was created, after much discussion and collaboration with key DH training communities and partners across the country, including the Canadian Society of Digital Humanities (a society that has existed in some form since 1986!) and formerly, Compute Canada. cc:DH/HN was the joint effort of several partner organizations and people, including ACENET, Calcul Quebec, and the SFU Research Computing Group. Led by Laura Estill, Canada Research Chair in Digital Humanities at St Francis Xavier University, it is a mechanism that allows students, faculty, librarians, and the public to receive crucial, hands-on training in computer-aided disciplines in the humanities. The work has been quietly operating as a mostly community-supported program, with multiple staff in the Alliance distributed workforce who contribute to its ongoing development by serving on the steering committee or acting as institutional liaisons. Its significance as a training resource for DH skills cannot be overstated as a vital and impactful part of a robust DH training network that already exists across Canada. This lightning talk will showcase how regional organizations and host sites participate in the Canadian Certificate for Digital Humanities as part of their support for DRI training nationwide.

We present an openly available data deposit and curation training course for researchers and research staff, composed of six self-directed modules that address all aspects of the data deposit and curation process. The course will bring a learner with no experience through a complete curation training journey. By completing the course, learners will earn a certificate of completion from Compute Ontario and gain the knowledge and skills to prepare FAIR datasets for deposit in an institutional data repository. Over the past decade, requirements and expectations for data sharing and deposit have grown tremendously, in part as a response to the 'reproducibility' crisis. Since September 2016, total deposited datasets in Borealis have grown by 172% annually, and users have risen by 154% annually. When the Tri-Agency requirement for data deposit phases in, we expect the volume of data deposits at Canadian institutional repositories will further increase. To ensure that data deposits follow the FAIR principles, and to prevent the disclosure of sensitive information, most Canadian institutional repositories have implemented a staff-mediated curation and review process. While the curation process is essential to ensure data quality, especially when researchers are untrained, it requires significant staff time and expertise. Without additional funding for more curators, an alternative model is needed. Our approach empowers researchers and enables institutions to maintain a consistent standard for data deposit with a more scalable curation model. Institutions can implement a local version of the course and if desired can recognize researchers who have obtained the certificate and allow them to curate their own datasets and deposit directly, bypassing the usual institutional curation process. This enables researchers and research staff to deposit data more efficiently and supports institutional capacity as the volume of data deposits continues to increase.

Ensuring high hardware utilization is a persistent challenge for High-Performance Computing (HPC) clusters. GPUs, in particular, are difficult to schedule efficiently; a single GPU may be too large for a given job, or the barrier to entry for researchers to leverage the hardware effectively is simply too high. While NVIDIA's Multi-Instance GPU (MIG) feature offers a solution, it introduces its own drawbacks, including extensive downtime for node reconfiguration and rigid restrictions based on fixed hardware configurations. To overcome these limitations, Vulcan system administrators Rahim and Karim developed SoftMIG. Inspired by Project HAMI, SoftMIG provides a flexible, software-defined approach to partitioning GPU resources. This talk will explore the technical details of SoftMIG's implementation, compare its capabilities directly to standard MIG, and demonstrate how it can be utilized to significantly increase GPU utilization across the Alliance HPC clusters.

While SD4H is designed to enable data-intensive health research across institutional boundaries, its effectiveness depends not only on technical safeguards and regulatory compliance, but also on a largely invisible layer of socio-technical governance work. This talk examines how this kind of community cloud, when integrated with shared national services (e.g., monitoring, centralized support, and common infrastructures), reshape privacy and security governance practices. Governance is increasingly distributed across institutional and infrastructural levels, creating both opportunities for harmonization and new operational complexities. In this context, data stewards, platform operators, and support teams play a critical role in mediating access, interpreting policies, and managing risk in real time. We highlight a central tension: national initiatives seek to accelerate research through standardized and scalable access to sensitive health data, while institutions remain accountable for privacy, security, and ethical oversight. This misalignment creates governance gaps, where responsibilities are shared but not always clearly coordinated, placing significant burden on intermediary actors. Drawing on emerging SD4H practices, we explore how infrastructure-level services redistribute and partially streamline governance work, while raising new questions around accountability and sustainability. By foregrounding the invisible labor underpinning secure data access, this talk advocates for interoperable, risk-based approaches that position infrastructure and platforms like SD4H as multi-level governance ecosystems enabling both innovation and trust.

Canadian research institutions are central to the digital infrastructure and services that support and empower researchers in an increasingly data-driven environment where effective and responsible data management is essential to research excellence. To continue understanding and monitoring the shifting landscape, the Research Intelligence Expert Group (RIEG) conducted a second national survey examining how institutions are building capacity for Research Data Management (RDM) and responding to evolving policy requirements. Administered in Fall 2025, the bilingual survey gathered insights from universities, colleges, and research organizations across Canada. The survey explored the impacts of institutions developing and implementing their RDM strategies in alignment with the Tri-Agency RDM Policy, including internal coordination, external collaboration, and the provision of local support for researchers. With 55 responses nationwide, the findings highlight both progress and persistent challenges in developing sustainable RDM capacity. Key themes include the need for highly qualified personnel, addressing critical skill gaps, investing in scalable and interoperable infrastructure, and expanded services to support data stewardship, access, and reuse. This presentation will provide a high-level synthesis of these findings, connecting institutional experiences with broader trends shaping Canada's research data ecosystem. It will emphasize the importance of shared services, partnerships, and coordinated strategies in advancing national capacity. By situating these insights within the broader Digital Research Infrastructure and services landscape, the session aims to inform planning, foster collaboration, and support the continued development of a more connected and resilient research data community in Canada.

How do academic support staff learn about their roles and responsibilities in the administration of the Tri-Agency Research Data Management Policy? Opportunities to learn about RDM are readily available for researchers and academic staff, but support staff are often the ones responsible for handling information, guidance, and implementation of funder policies. This lightning talk will introduce an upcoming SSHRC Connection Grant-funded event that will provide RDM training for people in research support roles in the Atlantic provinces. Institutions in Atlantic Canada are smaller, geographically dispersed, and often serve diverse research communities. Through a mix of presentations, panels, and hand-on activities, attendees will learn about RDM and participate in the creation of tailored resources that will support inter-departmental and cross-institutional collaboration. The aim of this event is to create adaptable and adoptable research data service models and create a community of practice that supports RDM in the Atlantic region. In this short talk, we will describe our approach to program planning and our vision for extending this event model to other regions in Canada. Sustainable research data support requires buy-in from everyone; with this talk, we invite reflection on how RDM training can be tailored to meet the diverse needs of our national research community.

The Sensitive Data Expert Group released the first iteration of the Sensitive Data Toolkit in 2020 to support researchers working with sensitive data in navigating evolving research data management (RDM) practices. A comprehensive refresh has since been undertaken to reflect developments in the field. The refreshed Sensitive Data Toolkit (2.0) offers an updated and expanded Glossary of Terms, Research Data Risk Matrix, and informed consent language, addressing data use, and storage. A new section titled the De-identification Guidance was added to the toolkit that provides guidance and recommends tools: sdcMicro and Amnesia, to minimize disclosure risk when sharing data from human participants.

Beyond the Badge: ACENET's Approach to Microcredentials
Grace Fishbein

The 2026 RDM Jumpstart
Nick Rochlin, Jennifer Abel, Maria Amoros-Teijeiro, Diana Bertuol Garcia, Jamie Clarke, Anneliese Eber, Jaime Orr, Sandra Sawchuk, Mathew Vis-Dunbar

Advancing Training Materials Discovery: What's New in Explora Phase II
Tania Tan, Ramses van Zon, Chris Want

Introducing Evolo: Calcul Québec's Learning Platform
Maxime Boissonneault, Caroline Baril, Moïra Dion, Cindy Marchi, Valérie Harvey

Designing Scalable Asynchronous Training: Insights from ACENET's Adoption of SCORM with Gomo
Tannia Chevez

The International HPC Summer School
Ramses van Zon, Tania Tan

"Running Kubernetes" can mean a lot of different things. A single-tenant bare-metal cluster dedicated to one project looks nothing like a shared multi-tenant environment where dozens of teams deploy into namespaces on the same control plane. Each model comes with trade-offs in isolation, resource efficiency, operational complexity, and security - and for HPC and systems teams evaluating Kubernetes or already running it, understanding those trade-offs matters before you commit to an architecture. This workshop breaks down Kubernetes tenancy models in the context of research infrastructure. We walk through hard isolation boundaries - separate clusters, dedicated nodes, VMs - versus soft boundaries built on namespaces, RBAC, network policies, and policy engines like Kyverno. We will talk about where namespace-level multi-tenancy works well, where it starts to break down, and how to decide when a workload should graduate to its own cluster. Attendees should walk away with a practical framework for evaluating tenancy models against their own institutional needs, a clearer sense of soft versus hard security boundaries, and concrete patterns from a production Kubernetes research platform.

Artificial intelligence (AI) has changed the way we work in a relatively short amount of time. All workflows stand to benefit from the inclusion of new technologies, but this must be balanced with the necessities of professional responsibility. Within research data management (RDM) this balancing act requires collaboration across local, regional, and national levels. This session creates an interactive environment in which RDM practitioners, DRI service providers, Network of Expert members, and interested members of the DRI community can collaborate to generate insights into the future of RDM with respect to AI. This session will invite attendees to consider how AI can be implemented now and in the future in RDM to facilitate workflows, while ensuring that it is done in a way that suits the needs of researchers and those who support them best. It looks to the future and asks how AI might be implemented in back-end systems and front-end, researcher-facing applications to better facilitate outcomes that encourage, support, and expedite outcomes in RDM. Outcomes from this session will help to identify gaps in existing knowledge and practice, as well as work toward facilitating acceptable use of these technologies, which will be used to inform future activities within the RDM Network of Experts and beyond.

README files play a critical role in enabling the understanding and reuse of research data, yet their inclusion and content remain inconsistent across Borealis deposits. In response to a community-identified need, the Borealis README Working Group (under the Borealis Expert Group) was established to align practices and improve documentation standards across institutions. This group brings together institutional perspectives to define shared expectations for README content that support more consistent and scalable curation and improve the usability and FAIRness of deposited data. Through an environmental scan and comparative analysis of existing README templates and elements, the group developed guidance establishing the minimum and enhanced metadata elements that should be included in a README for Borealis deposits. Building on this foundation, the group is developing an approach to integrate README creation into the Borealis deposit process. This includes conceptualizing a tool that structures and reuses metadata to generate an editable README that can be added to datasets at the time of deposit in Borealis. This approach aims to improve documentation quality at submission and reduce the need for intervention during curation. We will present our approach, key outputs, and next steps, demonstrating how coordinated community-led efforts can strengthen repository practices, build institutional capacity to support data deposits, and enhance the reusability of research data.

This presentation takes a look at the upcoming work on DMP Assistant. The Service Team have recently committed to an ambitious work plan that will take data management planning with DMP Assistant to the next level. This presentation will introduce new API connectivity enabled through the v2 API, versioning and publication workflows soon to be available to users, and forthcoming work on a new platform for DMP monitoring that will appeal to RDM administrators and others who might find it useful to compare contents of plans with other data sources. This presentation will be delivered by the Product Lead for DMP at the Alliance, with members of the service team (including developers and content staff) to answer questions as DMP Assistant prepares for its next major evolution.

Our account, access, and resource allocation mechanisms have evolved organically over many years and no longer align with contemporary business, research and operational needs. The current model relies on multiple parallel access pathways and a PI‑centric structure that introduces operational complexity, scalability limits, and growing strain on technical staff and review processes. Rising demand, persistent resource imbalances, and reliance on manual oversight have increased fragility and institutional risk, while adding new infrastructure continually amplifies structural weaknesses. Non‑expiring allocations, unclear lifecycle management, and tightly coupled identity and access systems further constrain adaptability. At the same time, there is a clear opportunity to modernize by rethinking eligibility, exploring project‑based allocation, clarifying policies, and separating scientific intent from infrastructure capacity decisions. The aim of the proposed workshop is to identify problems through staff expertise, with the aim of generating shared understanding and pragmatic solution directions. Ultimately, the goal is to inform a structured redesign that delivers a more scalable, equitable, and future‑ready system capable of supporting evolving computational research needs.

Érudit (https://www.erudit.org/) is a Canadian digital platform dedicated to disseminating scholarly research in the humanities and social sciences. Presenting over 170,000 articles from more than 300 Canadian journals, it plays a key role in advancing open science by supporting open access publishing models and promoting broad access to research outputs. As the platform continues to grow with an increasing number of articles and journals being hosted, there is a rising need to improve content discoverability, enhance user navigation, and better showcase the high quality of its corpus. This is particularly critical for science published in French, which, in the Canadian context, deals with locally and regionally impactful research topics, but which often lacks the high visibility and international circulation of English publications. Given that most of the Érudit's content is text-based, large language models offer promising opportunities for innovation, including semantic search, on-the-fly translation, metadata extraction or the training of domain-specific language models. These developments, however, must be guided by clearly identified needs and grounded in strong ethical principles, particularly in the context of large-scale data harvesting from open platforms by commercial actors. In this lightning talk, we will present our approach and share our vision for the next generation of the scholarly platform, one built for the common good, grounded in openness, and powered by collaboration and innovation.

This is a basic overview of what the Digital Research Alliance of Canada. We explain the resources, services, concepts, and terminology used, as well as the processes by which ARC resources move from allocation to use in science. This overview is for staff in Research Data Management (RDM), Research Software, DRAC Management, other non-Advanced Research Computing (ARC) roles, and new DRAC staff.

Persistent identifiers (PIDs) are unique identifiers for and long-lasting links to research "entities" (e.g. researchers, institutions, and publications). PIDs have metadata describing the entities they identify and are indexed and interlinked so that interested parties can query the PID ecosystem to answer questions about research activities. PID registries are community-governed sources of information operated by non-profit organizations. PID maintainers provide APIs, allowing interoperability with their systems. The Alliance supports PIDs as an RDM service, and there is an established culture of using PIDs in academic publishing (major publishers register DOIs for their papers and collect ORCID iDs to identify authors) and data repositories. That said, there is an opportunity for PIDs to solve problems in other parts of the research enterprise, including ARC. This workshop aims to identify short- and longer-term problems in the Canadian DRI ecosystem that PIDs could help solve. Research Activity Identifiers (RAiD) for projects are being piloted in Canada, so project-related use cases would be especially valuable. the facilitators will propose ideas for feedback and solicit other ideas from participants. Those ideas include: If RAiDs identify projects for which compute is awarded, those RAiDs could be used to assist in impact reporting by listing all the outputs generated by a project after compute was allocated for that project or otherwise tracking that project across its entire lifecycle. The Canadian Common CV (CCV) has been deprecated for use in the Resource Allocation Competition. Instead, CV information could instead be drawn from applicants' ORCID metadata (optionally, given ORCID data residency in the US). Similarly, a project's RAiD could be used to prepopulate research outlines. Prospectively, PIDs could be used to identify clusters or compute allocations. PIDs for research cores, instruments, and tools are an active area of development, and ARC system maintainers could contribute to that development.

T-CAIREM (the Temerty Centre for Artificial Intelligence Research and Education in Medicine) at the University of Toronto has developed Health Data Nexus, a platform for safe and secure data storage, streamlined access, and scalable compute and analysis. In this session, we will discuss the development of Health Data Nexus and its role in the research data management ecosystem, focusing on the pillars of research software and education. We will present two case studies of successful work using Health Data Nexus. First, we will discuss how the platform can play a role in student and trainee development through workshops and Datathons. Then, we will talk about research software using Health Data Nexus as a major part of its workflow through the example of automatic chart abstraction of sudden cardiac arrest patients. We will walk through the workflow for users to be onboarded onto the platform and become credentialed, allowing interested participants to work towards accessing platform data on their own. We will conclude with a look at the future direction of Health Data Nexus, the importance of data sovereignty, and the strong role that the Alliance can play in moving the platform forward.

Members of the Canadian Digital Research Infrastructure (DRI) ecosystem are regularly challenged by having to make decisions without the right operational data to inform those decisions. The original 2020 #frak report, which focused on HPC resource allocation, noted this systemic gap, flagging that the national ecosystem lacked the data environment required to optimally manage the infrastructure. Now, as the ecosystem grows in complexity, gaps between where data is, where it needs to be, and what it needs to look like when it gets there stand to have even greater impacts on operations. We are launching a comprehensive, year-long consultation process aimed at identifying the gaps in our operational data sharing environment. This project-#frak2-is a vital, community-driven follow-up to the first #frak that will produce a national roadmap for the future collection and dissemination of DRI operational data. This initiative is crucial for supporting experts across Advanced Research Computing, Research Data Management, and Research Software, as the resulting recommendations will define the operational data sharing standards necessary for effective system design, strategic planning, risk analysis, security, funding, and RDM integration, all while supporting researchers with real-time operational insights. Join us in this high-impact, 30-minute workshop to kickstart the #frak2 consultation process. We will provide a very short, targeted introduction to the core questions of the review and then immediately transition into a series of focused, topic-based roundtable discussions. This is an immediate opportunity to contribute your initial, expert thoughts on how to shape this review of the operational data needs of the Canadian DRI community.

The Curation Services team at the Alliance is developing a plan to expand research data curation support into the active phase of research, as the Active Data Management Service (ADMS) is procured and developed. To that end, we need to consider what we mean by "curating research data in the active phase" - what does curation mean? What is the "active phase" of research? We also need to consider what is already being done by others in the DRI ecosystem, to prevent duplication of effort and identify gaps where we can best improve researchers' efforts to improve reusability and reproducibility in their datasets. In this session, we will host a guided discussion about (1) what different groups mean when they use terms like "active data management", "research data management", "active research phase", and "data curation"; (2) who in research institutions support research data management already during more active phases of a research project; and (3) where are gaps in existing services that could be filled, and what could those look like?

Research generates an enormous amount of information about scholarly activities distributed across many different software systems. Persistent Identifiers (PIDs) - long-lasting digital references to and structured descriptions of People (e.g., researchers), Places (e.g., universities), and Publications - interlink this information, enabling the long-term findability, accessibility, interoperation, and reuse of academic work. PIDs are foundational for DRI, having become especially important for repository, publication, and research information workflows. However, they are not just technical plumbing, they are community-driven infrastructure, sustained through shared practices, governance, and coordination across the research ecosystem. PIDs do more than connect data: they actively shape scholarly workflows, recognition, and the persistence of trusted knowledge. Canada has become a global leader in the implementation of PIDs, especially given the central engagement of all sectors of research and their representation on the Canadian PID Advisory Committee (CPIDAC). The CPIDAC, with CRKN and the Alliance, just published Canada's National PID Strategy. National support, guidance, and funding for PID initiatives has and will continue to benefit Canadian research, and supports institutions in meeting their goals for research intelligence, open science, and to reduce administrative burden on scholars. In this session, John Aspler (Manager, Canadian PID Program, CRKN), Tristan Kuehn (Product Lead, Discovery Services, the Alliance), Dominique Roche (Senior Policy Advisor, SSHRC), and Alyssa Arbuckle (Strategic Research and Partnerships Advisor, CRKN) will highlight the value of PIDs for Canadian research and DRI services by presenting case studies - like the integration of PIDs into national funder systems - and by describing how the National PID Strategy encourages value-adding adoption of PIDs. This session will focus on concrete use-cases, like the implementation of PIDs and PID workflows for Awards, to demonstrate the value of PIDs to the broader research ecosystem. Moreover, it will contextualize PID activity within broader, national, and disciplinary-specific research infrastructure.

This Borealis Community Open House will provide a chance for Borealis community members to meet in person at DRI Connect in a dynamic, in-person version of the bilingual monthly Borealis community meeting. The session will also provide an overview of the Borealis data repository and associated community of practice for anyone who wants to learn more about them The Borealis Dataverse Repository provides a bilingual data repository service to more than eighty Canadian research institutions. A single instance of the Dataverse software application is hosted at the University of Toronto and supported by a small service team there. The community of administrators from participating institutions interacts with the service team through a listserv, support tickets, and monthly bilingual community meetings, which feature updates about the service and community knowledge sharing. Community members collaborate on projects of common interest through the Borealis Expert Group and its working groups. The expert group also serves as the Expert Advisory Council to the Steering Committee that governs the Borealis service. Some current community activities include: a major Dataverse software upgrade later this year, testing an integration of the Traditional Knowledge Label system into Borealis, collaborative development of a tool to generate README files to accompany data deposits, increased emphasis on bilingual support, and ongoing support for new joiners. This session will feature a short program of information from the Borealis Service Team, the Borealis Community Facilitation Team, and the Borealis Expert Group, followed by time for questions and unstructured discussions, just like a monthly community meeting.

Liquid immersion cooling is rapidly moving from niche curiosity to viable mainstream option for research computing. This talk walks through our recent deployment of a liquid immersion cooled HPC cluster - from the initial design decisions and hardware selection, through the physical installation, to day-to-day operation. We'll share what drove the move away from traditional air cooling, the practical challenges we ran into along the way, and the efficiency, density, and sustainability gains we've observed. Attendees will come away with a clearer picture of what liquid immersion looks like in practice, what to watch out for when planning a similar deployment, and where this cooling approach fits into the broader future of HPC infrastructure.

As ARC specialists, we often focus on "catching the bad guys". We identify wasted resources and punish extremely inefficient usage patterns. We spend disproportionately less time examining users who do not fall into the worst-performing percentiles. To address this gap, our project aims to capture the evolution of a single user's interaction with the system and gain a better understanding of usage patterns across all phases of the project lifecycle. To do so, we are tracking users over the course of short, finite pilot projects. Taking a closer look at what defines an "average" pipeline should benefit us in two ways: 1) by providing valuable insight into workloads that run comparatively smoothly and 2) by quickly addressing challenges faced by users who operate within expected norms. We present the data collected in 2025, during an AI project with an industrial partner from South Korea, as well as our planned approach for automating data collection and reporting in 2026.

Discover how we've rolled out suballocations in RAC 2026! Join us as we explore the reasons behind suballocations. Learn to identify when a group can benefit from the suballocation solution and how to request it. Watch as we walk you through the process of implementing a suballocation.

Imagine a major, transformational technology that appears seemlingly overnight and rapidly grows to dominate the conversation in every field, offering potentially massive but hard-to-measure efficiency and cost improvements, but demanding a complete re-thinking of every aspect of traditional information systems. Imagine that this technology spurs gigantic investment and interest, creating a whole new market that disrupts the traditional technology landscape, inspiring almost giddy optimism coupled with an overriding fear of missing out among leaders both within and far beyond the technology sector. We're talking about AI, right? Well yes, but as it turns out, we've been here before. The lifetime of cloud tech since the mid-2000s offers a portrait of a very similar phenomenon, and potentially illuminates a great deal of what may be ahead for AI - but with the benefit of hindsight! Let's look back at where we've been in cloud to see where we may be going with AI, and to maybe get a bit of a head start preparing for the coming AI future.

In this panel, we will introduce the National Data Spaces Pilot Participants and their plans for creating Canada's first official research-focused and community-centric data spaces. Through this Pilot, the Alliance will support the Pan-Canadian Genome Library, Canadian Integrated Ocean Observing System, Canadian Research Data Centre Network, and CanAvian in establishing their discipline-specific data spaces. This session will explore the Participants' current aims, anticipated challenges, and expectations for working with the Alliance and each other through the Pilot. We will discuss integration opportunities with the broader DRI ecosystem, including where National Data Spaces can work with national DRI services to enable better access to and coordination with discipline-appropriate research environments, services, compute and storage infrastructure. This will be an interactive session with questions from the audience. Data spaces are being introduced globally to address challenges of data access and trust. They provide an abstract layer where data providers and users can securely interact in a trusted environment to negotiate data access via machine actionable data policies. The data space model addresses researcher and data professional challenges of identifying trustworthy sources of data, especially for discipline-specific data that may not be appropriate or discoverable through generalist repositories, and; time requirements for manually negotiating data access and use. However, current data space models assume a high level of maturity and cohesion in communities who build data spaces, with agreed upon standards, policy requirements, and governance structures. The purpose of the National Data Spaces Pilot is to provide support for research communities to advance their maturity towards creating a data space while also identifying what a National Data Space should and can be for academic research communities in Canada.

HPC clusters are increasingly being used by researchers who didn't grow up on the command line. Agentic AI tools like Claude Code, Cursor CLI, and Continue CLI are showing up everywhere - and we wanted to see what that actually looks like on a real national HPC cluster. This talk is a show-and-tell: we'll walk through how users are using these tools today to write, submit, and iterate on SLURM jobs in plain English, and we'll demo both public and locally-hosted models to touch on data sovereignty. No grand conclusions, no framework proposals - just a live look at where things are headed. Come watch us do weird things with AI on a cluster.

This technical talk presents the Distributed Storage and Compute Grid (DSCG), a new set of infrastructure the Digital Research Alliance of Canada is building from bare metal up across three clusters and seven sites — two stretched (Alberta, Quebec) and one traditional (Ontario). We'll walk through the architectural principles, the platform model, and the technology choices behind the stack: bare-metal Ceph and Kubernetes, failure domains, RGW multisite replication, GitOps, and the observability stack. Attendees will leave with a concrete picture of how 16,896 CPU threads, 96 GPUs, and 100 of petabytes(raw) of object storage are being used to provide researchers with more cloud native infrastructure.

Bioinformatics researchers rely heavily on national HPC infrastructure, but effective resource utilization remains challenging because of diverse software stacks, evolving workflows, and varying user experience. As the Bioinformatics National Team, we are developing a coordinated strategy to improve how bioinformatics workloads use DRAC resources through three activities: workload analysis, training, and targeted consultation. First, we analyze ticket trends and heavily used software modules to identify common tools and workflows, then prioritize benchmarking and scaling studies for representative applications. So far, we have completed work on BLASTp, FreeSurfer, Kraken2, and Cell Ranger, covering major bioinformatics use cases in sequence alignment, MRI imaging, metagenomics, and single-cell RNA-seq. Second, we are strengthening training capacity by developing a shared slide deck that can be reused across institutions. We also collaborate with Canadian Bioinformatics Workshops, which will use DRAC Magic Castle to teach researchers how to run bioinformatics analyses in HPC environments. Third, we are expanding consultation efforts by comparing CPU and memory utilization across bioinformatics and non-bioinformatics groups to identify underused allocations and support right-sizing. We also collaborate with C3G on longer-term researcher support. This presentation will share our current strategy, progress to date, and the direction of this ongoing work, and will also provide an opportunity to gather feedback and suggestions from the community.

Institutions that are independently managed and have varying policies, as well as regional cultures and mandates, are often stuck developing guidance and tools on their own. This session will provide a knowledge sharing exploration of a simple and easily replicable tool that has potential to be adapted across institutions, taking into account their local context and needs. McGill's Claire Burrows and Lina Harper from McGill's Calcul Quebec office, will pair up to share institutional wisdom about a newly launched internal portal to help guide researchers through McGill's Cloud Directive. Researchers aren't always aware of data governance policies at their institutions. Cloud directives can be complex, technical, ambiguous. This year, McGill's Digital Research Services team launched a Sharepoint portal to translate some of those more mealy institutional data policies into intuitive guidance. DRS developed recommendations for selecting cloud-based research software using an intuitive traffic light framework: green for acceptable use, yellow for proceed with caution, red for consider other options. Relevance to the DRI: Responsible data management and cloud use sits at the intersection of data and people - which is the core of our work in the DRI. Aim: This lightning talk/fireside chat aims to be a knowledge-sharing exercise. We will present a replicable tool that other institutions could adapt for their local context: an internal knowledge base to help researchers assess which cloud-based research software might work for their individual projects, taking into account data sensitivity and local policies. Format: A 10 minute lightning talk/fireside chat, ending with a live short walk through & use case. We will open it up for discussion about how other institutions handle similar gaps - because of time limits, this discussion would be started on the

As open science mandates, funder requirements, and publisher expectations continue to evolve, research institutions increasingly face the challenge of enabling data sharing while responsibly managing research data that cannot be shared openly. Controlled‑access research data such as sensitive or restricted access data require appropriate governance and secure infrastructure to support access, reuse, and long‑term stewardship. The Controlled Access Management for Research Data (CAM) Initiative is a national effort led by the Digital Research Alliance of Canada in collaboration with Partner Organizations from across the Canadian research ecosystem. The initiative provides a forum for collaboration among research institutions, data repositories, and other stakeholders, to advance Canada's capacity to manage controlled‑access research data. The CAM initiative recognises that effective controlled access management requires alignment of infrastructure and technologies with data governance frameworks. Through working groups and pilot projects, CAM Partner Organizations collaborate to develop resources and test technologies which address challenges in facilitating data sharing in a way that aligns with ethical, privacy and security requirements. Previously, CAM Partner Organizations co-produced the CAM Report & Recommendations, which describes the current state, outlines CAM-related needs and challenges, and advances 16 recommendations regarding controlled access management for research data in Canada. Building on this foundation, the initiative supports ongoing working groups and activities focused on clinical data deposit, pre-deposit activities, the continuation of the FRDR Sensitive Data Project, and advancing the recommendations outlined in the report. This session will provide an overview of the CAM Initiative's aims, structure, and highlights how national collaboration can help institutions meet open science requirements.

Research teams are building more software than ever, web apps, data portals, analytical tools, and many of them end up deployed on standalone VMs or squeezed into HPC environments that weren't designed for long-running services. It works, but teams end up spending a lot of their time on operational plumbing instead of their actual research. Our team at the University of Victoria has built a managed multi-tenant Kubernetes platform on Arbutus Cloud to change that. Teams can get their own namespaces on a shared cluster with access to production-grade infrastructure, GitOps deployments, automatic TLS, monitoring, network policies, all managed by us. Bring your containers and you've got a production service without needing to become a Kubernetes expert. The DMP Assistant and Mountain Legacy Project have already moved onto the platform from conventional VM deployments. This talk walks through what we built, what it takes to get started, and why your configurations stay portable if you ever outgrow the shared environment.

During DRI Connect 2025 Jeff Albert and me have a presentation titled: "The future of HPC in the age of AI, AI clusters vs HPC systems. Are we ready" I think it would be interesting to have a panel discussion following that topic, especially that we have just completed the infrastructure refresh last summer.

Our team has been working to improve the CCDB, the Alliance's platform for accounts, allocations & more, by upgrading the framework (Rails) and language (Ruby) from versions that predated the iPhone 5. We share things we wish we'd known at the start: strategies that worked well, potential pitfalls, and principles that can help guide projects toward long-term maintainability. The goal is to be useful both to teams currently maintaining mature software projects and to teams building something they hope will still be running years from now.

The Arbutus Cloud, operated by Research Computing Services (RCS) at the University of Victoria, is a production OpenStack environment serving researchers across Canada. As a national resource, its day-to-day operation and user support increasingly depend on staff distributed across multiple institutions - including Simon Fraser University, the University of British Columbia, the University of Alberta, and the University of Manitoba. Historically, granting these staff administrative access required provisioning associate accounts through the University of Victoria's identity services, a process that was slow, difficult to audit, and disconnected from each individual's home institution identity. To address this, RCS adopted CILogon, a federated identity and access management platform operated by NCSA and the University of Chicago that enables authentication using existing institutional credentials. Its COmanage component provides collaborative organization management - allowing distributed teams to define enrollment workflows, manage group memberships, and release identity attributes as OIDC claims. CILogon's integration with the Canadian Access Federation makes it a natural fit for multi-institutional collaboration in the Canadian research computing community. Using COmanage enrollment flows, RCS assigns the Arbutus National Team role to approved accounts, and these roles are released as OIDC claims during authentication. Downstream, the claims are mapped to both Kubernetes RBAC roles on the underlying cluster and administrative roles within OpenStack, giving staff the appropriate level of access without any locally managed accounts. This talk will describe the architecture of this federated access model, the practical experience of configuring COmanage enrollment flows and OIDC claim mapping for a multi-tenant cloud platform, and the operational benefits realized - including simplified onboarding and offboarding, improved auditability, and a more sustainable model for nationally distributed cloud operations.

The SSCAN Data Space has been funded through the Alliance's National Data Spaces Pilot Funding Initiative. We aim to create a national infrastructure for social scientists working on sensitive data to work on and share their research data. This workshop is part of a much larger planned engagement with the research community to inform the SSCAN Data Space and will feed into our plans and development for the future. This workshop will begin with a very short presentation sharing our plans for the data space, and then move to a dialogue with the participants on our plans for its development over the next few years. SSCAN will have specific asks about what the community sees as essentials to support on the human side, the sorts of issues they face with curation of their current datasets that we should look forward to addressing, what the essentials are for the infrastructure itself, and how best to work with University Librarians and Data Support units to integrate a national infrastructure into their workflows. SSCAN also invites participants to ask questions and provide any other feedback on the project including any gaps that we haven't addressed that they would like to see us integrate. The results of the workshop will feed into a larger report on the results of SSCAN Data Space engagement sessions. Participants from all pillars are invited to participate and we welcome all perspectives -- though we expect RDM will be the most relevant stakeholder group for this workshop.