polly

Polymorphic Volume Scheduling

polly

Overview

Polly implements a centralized storage scheduling service that integrates with popular container schedulers of different application platforms for containerized workloads. It is an open source framework that supports use of external storage, with scheduled containerized workloads, at scale. It can be used to centralize the control of creating, mapping, snapshotting and deleting persistent data volumes on a multitude of storage platforms.

Key Features

What it does

Container runtime schedulers need to be integrated with every aspect of available hardware resources, including persistent storage. When requesting resources for an application the scheduler gets offers for CPU, RAM and disk.

To be able to offer persistent storage in a scalable way, the application and container scheduler needs awareness of the available resources from the underlying storage infrastructure.

Example workflow

  1. An application requires highly available storage with a specific set of policies applied
  2. The scheduler receives a request to start the application
  3. The scheduler checks with Polly or already has an off from Polly for storage resources
  4. Polly requests the volume(s) to be mapped to the container
  5. Scheduler issues request to start the container with persistent storage
  6. Container runtime orchestrates process of starting container and attaching persistent storage

Container runtime scheduler support

Cloud platform support

Storage platform support

libStorage

Polly makes use of the open source storage plugin framework libStorage to enable storage orchestrator tools and container runtimes to make requests of storage. Any storage platform that has a driver implementation for the libStorage framework will work with Polly.