Before adopting a new solution, Orca Studios ran a fully cloud-based infrastructure in AWS built around a self-managed ZFS storage server backed by Amazon EBS volumes. Storage usage averaged around 250 TB, but production cycles caused capacity to fluctuate between approximately 150 TB and 600 TB.

The studio typically operated between 40 and 110 virtual workstations, depending on the number of active projects. While the architecture performed well in earlier stages, scaling the system efficiently became more difficult as workloads increased.

During demanding workloads such as simulations and rendering, storage performance could become inconsistent. Burst performance from ST1-backed volumes worked well for short periods, but sustained activity could exhaust burst credits. When this occurred, throughput and IOPS dropped, and metadata operations slowed. These fluctuations affected version turnaround times and occasionally interrupted artists’ workflows.

Maintaining performance also required increasing operational attention from the IT team. ZFS-based architectures can deliver strong performance, but scaling and optimizing them often requires additional monitoring, tuning, and engineering work. For Orca Studios’ small infrastructure team, this added operational complexity at a time when production workloads were becoming more demanding.

At the same time, maintaining consistent performance often required moving to higher-performance EBS volumes and larger compute instances, which increased infrastructure costs. The team began looking for a solution that could deliver predictable performance while simplifying operations.

Difficult elastic scaling, interruptions to artists’ workflows, and increased costs were catalysts for finding a better solution.