If you’ve been working around HPC or distributed applications for any length of time, you’ve probably experienced the complexity and tedium of installing and configuring clusters. Whether on-premise or in the cloud you’ll find yourself installing nodes, configuring networks, and setting up shared storage, workload managers, and reporting tools; all before you even get to your application.
Grid Engine users can take heart. It’s now easier than ever to spin-up fully functional Univa Grid Engine (UGE) clusters on AWS Marketplace using a 1-click installation process. Whether you’re deploying clusters just for testing or running large-scale simulations, the automated installation process makes installing a cluster easier than ever.
Acknowledging the thousands of production deployments in enterprises of all sizes, and over 1,000 integrated applications, Univa Grid Engine supports the widest variety of distributed applications in fields like HPC, analytics, and machine learning / deep learning. By making scalable Grid Engine services available through AWS Marketplace, users can deploy clusters tailored to their specific application requirements. Users can select the appropriate Amazon Machine Images (AMIs) and amount of AWS Elastic Block Storage (EBS) depending on their requirements.
Unlike other cloud-based cluster solutions that deploy a full complement of nodes, the UGE service deploys with a single head-node and leverages built-in Navops Launch services (formerly Unicloud) to allow users to grow and shrink clusters with a single command. Clusters can also be configured to auto-scale based on workload demand using regular or spot instances. These features help maximize flexibility while keeping infrastructure costs to a minimum.
Depending on the AWS instance type and zone, Univa Grid Engine clusters deployed through the marketplace can easily scale to thousands of nodes supporting millions of jobs per day. Navops Launch functionality (pre-installed and configured on the marketplace deployed master node) enables “bring your own AMI” functionality allowing users to replicate on-premise environments in the cloud. This helps ensure application mobility and minimizes the chance of compatibility issues when running in the cloud.
After installing the master host via the marketplace, users simply ssh to the head node, scale their cluster as needed and start running jobs. Adding five nodes to the cluster is as easy as adding 500.
As workloads change, you can dynamically alter the instance types deployed to your cloud-based cluster.
Whether you’re a battle-hardened Linux admin or a first-time Grid Engine user, automated installation via the marketplace is something we can all get excited about.