OpenCue Sony Pictures Imageworks case study

How Sony Pictures Imageworks runs OpenCue in production

This page provides a case study of how Sony Pictures Imageworks runs OpenCue on production infrastructure. This case study illustrates an earlier version of OpenCue prior to open sourcing the project. This case study is aimed at system admins and other professionals planning to install OpenCue. When planning a production deployment of OpenCue, you can review this case study alongside the OpenCue getting started guide.

Before you begin

Many of the OpenCue terms and concepts in this case study are explained in more detail in the following introductory resources:

As you read through this case study, you might find it useful to refer to these introductory resources.

System components and specifications

The production deployment consists of the following components:

  • Several Cuebot virtual machines (VMs) servers
  • A database server that stores data over a Network File System (NFS)
  • A render farm consisting of between 2,500 and 4,000 render nodes, including:
    • Dedicated render nodes running RQD
    • Artist workstations, also running RQD
  • A 10 Gb/s network

From their workstations, artists submit jobs to OpenCue through a cluster of Cuebot servers. The Cuebot servers dispatch individual frames in a job to the render farm. Cuebot servers also store all persistent state and transactions in the database server. Figure 1 illustrates how the various OpenCue infrastructure components interact:

OpenCue infrastructure components

Figure 1. OpenCue infrastructure components

Each Cuebot VM is provisioned as follows:

  • Managed via vSphere platform
  • Allocated 4-core CPUs (Intel Xeon L5640 running at 2.3 GHz)
  • Allocated 8 GB RAM

The database server is provisioned as follows:

  • Runs on a bare metal server
  • Allocated 16-core CPUs (Intel Xeon E5-2670 running at 2.6 GHz)
  • Allocated 128 GB RAM

The back end storage is provisioned as follows:

  • Runs on a NetApp filer
  • Allocated the following types of storage:
    • Primarily using Serial Attached SCSI (SAS) 10K drives
    • Some SSD for caching
  • Connected to the database server over NFS

The current dataset occupies approximately 1.2 TB, including 7 years of historical data.

See also

To learn more about the production use of OpenCue, see the recording of the OpenCue Birds of a Feather roadmap from SIGGRAPH 2019. In this recording, Ben Dines from Sony Pictures Imageworks provides a summary of the development and use of OpenCue on a number of films.

What’s next?


Last modified Thursday, November 7, 2019: OpenCue Sony Pictures Imageworks case study (#111) (7b8ef28)