New Features in MMCloud 1.1 Release
Release Date
Released on 12-01-2022
Supported Clouds
MMCloud is
designed to work on any cloud infrastructure. Release 1.1 supports the following
clouds:
New Features
Platform
- Self-service license registration via the MMCloud license portal
- MMCloud CLI for submitting and managing workloads, and for maintaining the
MMCloud platform
- One-click download and installation of CLI binary for Linux, macOS, or Windows clients
- Web-based console for invoking CLI commands without needing to download the CLI binary
- User and group management — add, update, or delete users or groups
- Global login session (eliminates per job login)
- On-line MMCloud software upgrade using CLI commands
- Synchronization of CLI client version with MMCloud version
- Support for multiple data volume types, including local file system mount, EBS volume, and remote NFS mount
Cloud Resource Management
- Per job virtual machine (VM) allocation policy
- Customizable VM allocation policy, for example, try Spot Instance first, then On-demand instance
- Price limits for allocating Spot Instances
- Downloadable usage and cost reports
Security
- Extra security group attached per job
- Limited authority for IAM roles to provide job insulation
- Validation of important CLI commands such as "float submit job"
- Double confirmation on critical CLI commands, for example, when upgrading software or canceling jobs
- Selection of virtual private cloud (VPC) in which to deploy MMCloud (not limited to default VPC)
Container Images
- Built-in library of container images
- Image pull from popular container registries, such as DockerHub, GitHub Container Registry, quay.io (public repositories only), and Amazon Elastic Container Registry (public repositories only)
- Image upload from local server
Workload Management
- Runtime service configuration to customize execution environment
- Pre-defined environment variables for inclusion in job scripts
- Detailed information on running jobs available using CLI
- Real-time metrics for running jobs
- Autosave of stdout and stderr logs
- Customizable tags for each job, for example, for charge-back
- Port mapping between container and container host for interactive sessions
- Job migration between different VM configurations
- Automatic job migration for Spot Instances
- Support for Cromwell workflow manager
- Customizable filters to expedite job search
- Archival of completed jobs