Tortuga is a general-purpose cluster and cloud management framework with applicability to a broad set of applications including High Performance Computing, Big Data frameworks, Kubernetes and scale-out machine learning and deep learning environments. Tortuga automates the deployment of these clusters in local on-premise, cloud-based and hybrid-cloud configurations through repeatable templates.
Through automation, Tortuga greatly simplifies the creation of clusters; to demonstrate that, in 2016 Univa worked with Samsung SDS to validate this with Kubernetes. This test confirmed that Launch was 5x faster at installing large clusters and its automation reduced the number of steps by two-thirds.
At its most basic, Tortuga places and configures software on compute resources. Software and its related configuration information are packaged in “kits” (which look like a RPM with meta-information). Resources are abstracted through the use of adapters. Tortuga installs kits on the selected resource adapter.
The “cloud resource adapters” [also kits] that Tortuga currently supports are the following cloud/virtualization platforms:
For dedicated clusters in the cloud, Univa has released our AWS Marketplace offering that leverages Tortuga under the hood to make it simple to spin up and scale Grid Engine clusters in the cloud. You can read more here.
In a recent survey of Univa customers, of those ‘planning to deploy” or “deployed” in HPC loud, 70% indicated the intention of implementing a hybrid cloud which is a built-in capability of Tortuga.
Tortuga key Features:
Many enterprises have validated (and secure) images that they bring to the cloud and Tortuga has full support for bring-your-own-image (BYOI).
Made for enterprise integration
Tortuga was designed for enterprise use. Since no two enterprises run IT the same, or have the same network configuration or IT tool, Launch integrates well with existing “brownfield” corporate IT tools.
Then there’s the commercial product
For organizations experiencing increased volumes of high priority workloads who need the confidence of consistent, repeatable template-based cloud provisioning, metric and time-based policy automation and technical support, Navops Launch is our commercial offering. You can read more about Launch here.
Navops Launch Key Features:
Policy Management tied to Workload, not infrastructure
Scaling up or down of cloud-based infrastructures is dynamically managed according to workload metrics through use of a flexible policy engine. The built-in policy engine allows users to dynamically scale up or tear down cloud instances using Grid Engine workload metrics; in fact any metric.
In an upcoming blog we will look at using policy to dynamically manage HPC cloud instances.