High Performance Computing with Spot

Achieve The Most Efficient HPC Cluster on Earth

Boost your HPC performance while reducing costs by up to 90%

Integrate with HPC schedulers for the best possible mix of compute in the cloud

High-Performance Computing (HPC) allows scientists and engineers to solve complex, compute-intensive problems. HPC applications often require high network performance, fast storage, large amounts of memory, very high compute capabilities or all of these.

Spot enables customers to increase the speed of research and reduce time-to-results by running or bursting HPC to the cloud. Spot helps to reduce infrastructure costs by up to 90% using CPU, GPU, and FPGA servers from cloud excess capacity (aka “spot instances”), and reducing the overhead of managing the infrastructure, saving scientists and engineers time to focus on what’s next.

Spot’s Elastigroup product sits between the HPC Scheduler and your compute grid

Elastigroup deploys a controller (agent) that monitors HPC schedulers’ queues. You can configure Elastigroup to monitor specific job queues and Elastigroup finds jobs that are pending for execution, calculates the resources that these jobs need and launches the required compute to run these jobs.

 

 

Spot Gets Smarter

As you submit jobs and tasks, Elastigroup is learning what resources the tasks need and how long they will take. Elastigroup uses this information to densely pack tasks, without oversubscribing, to ensure fast and cost-effective runs. By combining task awareness (time & resources) with prediction of spot instance price & availability, Elastigroup is able to schedule tasks where they won’t be interrupted.

 

Simple to use

Simplify management & orchestration using Elastigroup – both provisioning & autoscaling of cloud infrastructure

Spot Instance Lifecycle Management

Elastigroup uses its proven prediction algorithm to match spot instances that can run for the desired amount of time for the required task or job.

Mixed Instance Types

By configuring heterogeneous clusters and mixing nodes of multiple instance types, Elastigroup delivers maximum efficiency to run the relevant job on the right machine type to achieve greater data processing performance.

Customer Spotlight

Demandbase

"We are running multiple of our production workloads on Elastigroup, across containers, Big Data and web services"
Michael Waltz | Principal DevOps Engineer, Demandbase

Optimizing for pricing model, instance type and size

Elastigroup automates the instance lifecycle and provides a combination of EC2 Spot, existing RI reservations and on-demand instances based on EC2 Spot capacity, availability & pricing trends

 

Mixed Instance Types

By configuring heterogeneous clusters and mixing nodes of multiple instance types, Elastigroup delivers maximum efficiency to run the relevant job on the right machine type to achieve greater data processing performance.

The World's Leader in management of Spot Instances

Elastigroup balances the required performance, cost, and SLA requirements when launching, scaling and terminating EC2 Spot Instances. By combining Elastigroup’s proven prediction algorithm for EC2 Spot availability.

Complete Confidence while running spot instances for production and time-sensitive jobs

Elastigroup drains nodes up to 15 minutes before a spot instance termination notification arrives from the cloud provider, gracefully terminating existing tasks and preventing the scheduling of new jobs on these nodes.

Integrated with Leading HPC Platforms

Elastigroup integrates with leading HPC platforms including the following:

  • Grid Engine & Univa Grid Engine
  • SLURM Workload Manager
  • PBS WORKS
  • IBM LSF – Workload Management
  • IBM Symphony
  • AWS Batch
  • Azure Batch