Overview

The ongoing revolution enabled via containerization, virtualization, and new orchestration models has dramatically changed how applications and services are delivered and managed across the computing industry. This revolution has established a new ecosystem of tools and techniques with new, flexible and agile approaches, and continues to gain traction in the HPC community. In addition to HPC-optimized container runtimes, emerging technologies like Kubernetes create a new set of opportunities and challenges. While adoption is growing, questions regarding best practices, foundational concepts, tools, and standards remain. Our goal is to promote the adoption of these tools and introspect the impact of this new ecosystem on HPC use cases. This workshop serves as a key venue for presenting late-breaking research, sharing experiences and best practices, and fostering collaboration in this field. Our fifth workshop iteration will continue to emphasize real-world experiences and challenges in adopting and optimizing these new approaches for HPC.

Scope

The scope of this workshop is to better understand and improve paradigms around containerization, virtualization and orchestration for HPC use cases. The most well-known approaches are containers and virtualization, but anything to further these goals is welcome. Topics include but are not limited to:

* Container runtimes, virtual machine implementations, and other related technologies
* Perspectives from container framework developers and maintainers
* System and architecture portability
* Scientific reproducibility, including FAIR considerations
* Experience reports from users, developers, and system administrators
* Lessons from exascale
* HPC in the cloud and/or cloud in the HPC
* GPUs, accelerators, proprietary interconnects, and other hardware considerations
* Security considerations when adopting container technologies, including zero trust models
* Container and virtual machine image management, including distribution and archiving
* Workflows, including interaction between traditional HPC applications and containerized workflows or edge services
* Performance and scaling studies with containers and/or VMs
* Orchestration, scheduling, and/or resource management of jobs and microservices
* New interaction techniques such as web apps (Jupyter, Rstudio, etc.)
* Community standards, including OCI
* Role of containerization in DevOps (ex: Gitlab CI/CD)
* Efforts to provide and curate containers
* Improving container user experience
* Future of containers (ex: web assembly, portability)
* Perspectives on container outreach- convincing native application users/devs to make the jump

Workshop format

Although in the past several years CANOPIE-HPC has followed a more traditional format, based on SC22 attendee feedback we plan to adjust the format for SC23 to include additional lightning talks, panels, and open discussion. The content of technical sessions will be driven by the mix of accepted submissions curated by the program chairs. We will offer two submission tracks: one for full paper submissions, and one for lightning talks.

Full paper submissions will receive at least 3 peer reviews in the SC23 submission system on the basis of scientific validity, impact to the field, reproducibility, and opportunity for interactive discussion at the workshop. The program committee will discuss the submissions and their reviews over a conference call moderated by the program chairs; final proceedings will be selected by majority vote of the committee. As in all previous years, accepted submissions will be published by the IEEE Computer Society for inclusion in the IEEE Digital Library. Artifacts for reproducibility (digital archives of code, README instructions, images, etc.) will be a significant factor in paper acceptance into the workshop. Further details are available in the “Submission procedure” section below.

For lightning talks, participants can submit an abstract that will be reviewed by the program committee. Our lightning talk track is meant to showcase demos, exploratory and late-breaking work, pain-points or shortcomings in current container/virtualization technologies, and projects that may not be fleshed-out enough to warrant a paper submission. We especially encourage lightning talk submissions from early career participants and people who are relatively new to the world of containers/virtualization.
Diversity and inclusivity

Similarly to the broader field of HPC, the organizers recognize concerns with diversity and inclusion in CANOPIE-HPC. We want to continue to improve diversity and inclusion at leadership, program committee, and participation levels. The expanded lightning talks session is intended to lower the barrier to entry in the workshop and involve more early-career and non-expert participants. We also plan to include an open discussion section to invite additional audience questions and participation. Virtual attendance (another opportunity for inclusion) will be guaranteed. We’ll make sure we have someone from the program community monitoring the chat throughout the workshop to address any issues or questions from our virtual attendees.

Submission procedure

Accepted manuscripts will be published in IEEE Computer Society.

Upload your paper in PDF format. Papers must use the new ACM proceedings templates (https://www.acm.org/publications/proceedings-template) and the CCS2012 guide. Manuscript length must be between 6 and 12 pages, including references and figures.

We enthusiastically welcome original, high-quality submissions within the scope above. These may describe complete studies; work-in-progress research; position papers on controversial, emerging, or hot topics; state of the practice; or any other manuscript the authors believe should be included in the CANOPIE-HPC program. We encourage submissions from academia, industry, government, and/or any other type of institution. Each manuscript will be assessed using peer review by program committee members (or outside reviewers, if needed) on the basis of scientific validity, impact to the field, reproducibility, inclusivity, and opportunity for useful and lively discussion at the workshop. Review will be single-masked; i.e., reviewers will know authors’ identities, but not vice versa. Authors should not anonymize their manuscripts. Each manuscript submission will be checked using anti-plagiarism software and can expect to have at least 3 reviews.

In conjunction with the SC Reproducibility Initiative, submissions should be as transparent as possible regarding all methods. Submissions are also expected to provide reproducibility artifacts such as reproduction instructions, source code, build recipes, and/or container images. These should be included as part of the submission in the form of Artifact Description/Artifact Evaluation appendices (https://sc23.supercomputing.org/program/papers/reproducibility-appendices-badges/), which adhere to an established format adopted by SC23.

The program committee will discuss the submissions and their reviews and select the program over a video conference meeting. Submissions will be assessed as-is, with no expectation of substantive revision after peer review, though some submissions may be accepted conditional on specified changes (“shepherded”).

Submit your SC23 CANOPIE-HPC paper: https://submissions.supercomputing.org/?page=Submit&id=SC23WorkshopCANOPIEHPCSubmission&site=sc23

Submit your SC23 CANOPIE-HPC lightning talk: https://submissions.supercomputing.org/?page=Submit&id=SC23WorkshopCANOPIEHPCLightningTalkSubmission&site=sc23