Remote Workstations for Discerning Artists
summary
Netflix builds cloud remote workstations — Netflix Workstations — using Spinnaker, Packer, Salt, EC2, DynamoDB, SQS and a Go agent to give artists one‑click seats. The team iterates between bake run acquire strategies to trade bake time for wait time, enforces 100 packages as immutable versions, and supports NICE DCV and Teradici remoting. Pools and role based access control exist, usability wins over cost, latency across regions remains a measurable problem, and the platform already works well enough to onboard distributed artists.
AI-written summary · about
chapters (15)
- 0:00Remote Workstations
- 1:36Technology
- 2:43What was the problem?
- 4:27Why not ship a computer?
- 7:05Who are Netflix Workstations for?
- 10:41How did we make Netflix Workstations?
- 13:10Components
- 14:39Management
- 17:36Configuration
- 19:48Package
- 27:11Remote Display Protocols
- 28:36User Access: RBAC
- 30:04Full Workflow
- 31:13What we learned
- 33:44What's next?
related talks
29mHow Netflix Thinks About DevOps
29mFast, Safe, and Continuous -- Picking all 3 at Netflix
27mBehind the Scenes: Running Critical Services @ Netflix
31mHow Netflix Thinks About DevOps
50mFast Delivery
31mKeeping Your Cloud Footprint in Check
35mPushing innovation and accessibility with real-time and the cloud
58mBuilding a Culture of Continuous Delivery