Carme (/ˈkɑːrmiː/ KAR-mee; Greek: Κάρμη) is a Jupiter moon, also giving the name for a Cluster of Jupiter moons (the carme group).
Or in our case:
an open source frame work to mange resources for multiple users running interactive jobs (e.g. Jupyter notebooks) on a Cluster of (GPU) compute nodes.
Follow us on Twitter : #OpenCarme
Combine established open source ML and DS tools with HPC back-ends
- Use containers -> Singularity
- Use Jupyter Notebooks as main web based GUI-Frontend
- All web front-end (OS independent, no installation on user side needed)
- Use HPC job management and scheduler -> SLURM
- Use HPC data I/O technology -> ITWM’s BeeGFS
- Use HPC maintenance and monitoring tools
- Open source
- Carme uses only opensource components that allow commercial usage
- Carme is open source, allowing commercial usage
- User Management
- User quotas (GPU time, priority, GPUs per job, jobs per time, Disk quota)
- Different User Roles (Quotas, right to add containers)
- Container Management
- Container store (user selects from predefined containers)
- Adding of user defined containers
- Scheduler
- Resource reservation (calender)
- Job queues for large jobs and instant interactive access for small jobs
- Data Management and I/O
- Redundant, global file system (BeeGFS), mounts into container
- Temporary job FS on local SSDs for max performance (BeeOND)
- Web-Interface
- HTTPS and SSH (if allowed) access via proxy
- Web front-end (management and IDE)
- The Carme prototype (beta) is currently up and running on our Cluster
- First public beta reales: February 2019
- First stable release: June 2019
- Carme is currently developed at the machine learning group of the Fraunhofer Competence Center HPC
- Contact: info@open-carme.org
We are open for contributions!