System Capacity and Sizing

Learn the default system capacity and scale parameters.

The following is the default system capacity and sizing.

  • Single AWS m7i family instance running Code Ocean core services, with m7i.large as the minimum requirement.

    • 100 GB EBS gp3 docker data volume

    • 500 GB EBS gp3 persistent data volume

  • Two AWS auto-scale groups (ASG) running user computations

    • General purpose r5d.4xlarge worker instances (configurable)

      • Instance specification:

        • 128 GB memory, 16 CPUs, 10 Gb Network

        • 300 GB EBS gp3 docker data volume

        • 300 GB Instance Store NVMe SSD working directory, with EBS autoscaling to attach additional EBS volumes as needed

      • Minimum ASG size: 0

      • Maximum ASG size: 10

    • GPU g4dn.4xlarge worker instances (configurable)

      • Instance specification:

        • Single NVIDIA T4 GPU

        • 64 GB memory, 4 CPUs, 25 Gb Network

        • 300 GB EBS gp3 docker data volume

        • 225 GB Instance Store NVMe SSD working directory, with EBS autoscaling to attach additional EBS volumes as needed

      • Minimum ASG size: 0

      • Maximum ASG size: 10

  • Flex compute resources available for user computations (configurable, adjusted to worker instance type)

    • General purpose

      • 1 core / 8 GB RAM

      • 2 cores / 16 GB RAM

      • 4 cores / 32 GB RAM

      • 8 cores / 64 GB RAM

      • 16 cores / 128 GB RAM

    • GPU

      • 1 GPU / 4 cores / 64 GB RAM

  • Advanced compute capsule resource management using dedicated machines*

  • Compute Capsule size limits:

    • Git repo size <5 GB

    • Git repo file size <100 MB

* The Code Ocean dedicated machine feature provides all the resources of an automatically managed EC2 instance of choice to run computations. Users can select the right computational environment needed to complete tasks that could not be achieved using standard flex resource classes.

Last updated