The most important items to think about:

• Are you looking to expand your current on-premise infrastructure into the cloud?
• Are you looking to move completely into the cloud?
• How are you going to connect your facility to the cloud? VPN or a direct fiber link?
• How are you going to manage your assets between on-premise and cloud?
• How to access the Houdini license server which lives either on premise or in the cloud?

Houdini License Server

The Houdini license server can live either on-premise at your facility or in the cloud. As long as the machine serving licenses or acting as a render node follows the system requirements.

By default your Houdini licenses are keyed to the following private networks:

Cloud Providers

• AWS Thinkbox
  
• GridMarkets

AWS Thinkbox

Once your rendering pipeline is integrated with AWS Thinkbox, you can scale rendering workloads to thousands, or even tens of thousands, of cores in minutes. You can also scale down just as quickly as you scale up, providing incredible compute elasticity and cost control.

• AWS Thinkbox Portal (small boutique studios)
• Direct Connect (larger studios)

AWS Thinkbox Portal creates a render farm, or bridges your current render farm, using AWS EC2 spare capacity, known as EC2 Spot Instance. The AWS Portal facilitates the secure communication between on-premise and AWS, handling asset transfer and software licensing. AWS Portal is the easiest way to get off the ground, since it automatically copies data from your on-premise storage to the cloud as needed. When you submit from Houdini with the Deadline plug-in, AWS grabs all the external file dependencies and uploads them to the cloud before the render (or simulation) starts. AWS Portal is a feature in Deadline and is a good tool for smaller studios.

To optimize the movement of data in production using the cloud, a dedicated network connection such as AWS Direct Connect (DX) is recommended to increase the available bandwidth and to decrease latency. There are solutions for scaling your direct network connection on short notice, up to 40GB using DX.

If you need hundreds or thousands of machines for your render farm, you may look to using EC2 Spot Instances for cost efficiency and filesystems like AWS FSx. You would need to copy your assets to the cloud with some sort of automated syncing, depending on what on-premise file server you are using. You can also sync data as you go using AWS DataSync, or if you have more time you can use AWS Snowball.

This Quick Start reference deployment guide provides step-by-step instructions for deploying an example Linux framework architecture for visual effects (VFX) burst rendering on the AWS Cloud by using the best practices and recommendations for content security adopted by the VFX industry.