CloudyCluster V3 Released
CloudyCluster version 3 was released and is available in the GCP Marketplace
We are excited for this release, we have spent a lot of tie with beta customers and extensive testing to get this new release to the marketplace. We are looking forward to extending the reach of investigation and research to the Cloud in a familiar campus / lab HPC environment.
The CloudyCluster team has been working hard, reducing latency, creating tighter integration with Google cloud, including the automatic use of placement policies for C2 instances.
Additionally CloudyCluster v3 provides:
- faster job response time
- better job error handling and reporting
- more robust limit and resource exhaustion notification
- enhanced GPU support and updated CUDA and relevant drivers.
- automatic GCS bucket creation for the HPC environment
- support for firestore native mode for the environment metadata.
- Linux kernel 5.4 support including page cache integration with OrangeFS
- Updated support for Globus
Security and encryption has been reviewed and improved to provide better piece of mind that your research data in CloudyCluster can be configured to meet the controls you need.
Billing related job directives that can support both Slurm or PBS jobs are available through CCQ (CloudyCluster's meta-scheduler), this makes it easy to crate labels in jobs that can tie back costs to users and research-projects.
CloudyCluster now natively provides support for stackdriver, making troubleshooting and performance tracking easeier across elastic compute nodes.
A wide variety of performance improvements, bug fixes and software updates are also part of this major release.