Young & Associates (YAGroup), an architecture and design firm with a nationwide team of visualization engineers, relied on an aging, hands-on on-premise render farm to turn out photorealistic 3D renders for client projects. ASCENDING partnered with YAGroup to migrate that rendering workload to AWS Deadline Cloud, giving designers elastic, on-demand render capacity without the burden of managing physical machines.
Background
YAGroup's design team works primarily in Autodesk 3ds Max with Chaos V-Ray for photorealistic rendering, producing visualizations for architectural and design projects. Historically, rendering ran on an on-premise farm managed through Deadline Worker and RPManager, with designers directly responsible for adding, removing, and troubleshooting render machines.
That model gave the team flexibility, but it also meant every capacity change or hardware issue landed on the designers themselves rather than on dedicated infrastructure staff, a growing burden as project volume and rendering complexity increased. YAGroup leadership, including team lead Daniel Simacek, engaged ASCENDING to plan and execute a move to a cloud-based rendering pipeline that could scale with demand and reduce the operational load on design staff.
The Challenge
Moving a live, production-critical render farm to the cloud without disrupting active design work required careful planning across infrastructure, workflow, licensing, and cost management.
- Zero-disruption cutover: Designers needed to keep submitting and completing render jobs throughout migration, with no interruption to active projects.
- Elastic capacity without cost overruns: Peak workloads required scaling up to match on-premise capacity (8+ render nodes), while idle periods needed to scale back down to avoid paying for unused compute.
- Tight integration with existing tools: The solution had to work with YAGroup's existing 3ds Max/V-Ray pipeline and their Egnyte network file share, which designers rely on to access and retrieve project assets.
- Operational handover: YAGroup needed a support model and documented playbook so their team (and ASCENDING's support engineers) could troubleshoot rendering issues, from failed jobs to license checkouts to network drive mount failures, without deep AWS expertise.
- Cost visibility: Leadership needed clear, predictable cost modeling before committing to a full production cutover.
Why ASCENDING
ASCENDING brought hands-on AWS Deadline Cloud migration expertise and a structured, phased migration methodology tailored to render farm workloads, where cloud infrastructure design must account for 3D rendering realities such as licensing, worker scaling, and asset syncing.
Rather than a disruptive lift-and-shift, ASCENDING recommended a rehosting strategy: stand up AWS Deadline Cloud alongside the existing on-premise farm, let designers transition gradually, and only disable on-premise access once the cloud environment was fully validated. This approach reduced risk while giving ASCENDING time to fine-tune scaling, licensing, and network configurations against real production jobs. ASCENDING also used AWS Migration Hub Application Discovery Agent up front to gather real on-premise infrastructure utilization data, ensuring AWS Deadline Cloud sizing and TCO estimates were grounded in observed usage rather than assumptions.
The Solution
ASCENDING designed and deployed an event-driven, auto-scaling rendering architecture on AWS Deadline Cloud that mirrors YAGroup's existing workflow while removing the manual overhead of managing physical render nodes.
- Assessment and sizing: Used AWS Migration Hub Application Discovery Agent to capture on-premise infrastructure utilization, informing a right-sized AWS Deadline Cloud environment (targeting 42 vCPUs / 90GB RAM peak capacity and 4TB of S3 storage).
- Environment setup: Stood up an AWS Deadline Cloud farm with IAM roles, security groups, cost controls (AWS Budgets and Cost Explorer), and an S3-backed storage layer for render assets and output.
- Event-driven scaling pipeline: Configured the Deadline Cloud queue to trigger an Amazon EventBridge rule whenever jobs arrive, invoking a Lambda function that sets the desired capacity of an EC2 Auto Scaling Group and dynamically launches worker instances from a custom AMI (deadline-fresh-image-v7) only when needed.
- Rendering plugin integration: Deployed the AWS Deadline Cloud submitter for 3ds Max so designers submit jobs the same way they always have, with V-Ray license checkout handled automatically through a dedicated license server instance.
- Asset and file integration: Built a MountNetworkDrive queue environment script so every worker automatically mounts YAGroup's Egnyte network share on boot, keeping the cloud pipeline consistent with designers' existing local-drive workflow; completed renders sync from S3 back to Egnyte for designers to retrieve.
- Testing and validation: Ran hybrid on-prem/cloud workflows in parallel, load-tested job queues, validated failover and recovery, and confirmed rendered output landed correctly in S3 before cutover.
- Cutover and rollback readiness: Executed a staged cutover by redirecting job submissions to the cloud, then disabling on-premise farm access, with a documented rollback plan in case critical issues emerged.
- Post-migration support and playbook: Delivered a 3-day hypercare support window plus a documented operational playbook (job-log debugging sequence, AMI update process, and a real-world example of diagnosing a failed network-drive mount) so YAGroup and ASCENDING support teams can maintain the environment going forward.
The Outcome
YAGroup now runs its rendering pipeline on a fully elastic, cloud-native foundation that scales automatically with demand and removes the day-to-day burden of physical render farm management from its design team.
- Elastic, event-driven scaling: Render capacity now scales from a single worker up to 8+ instances automatically based on queue depth, matching prior on-premise peak capacity without manual intervention.
- Predictable, modeled cloud costs: TCO planning showed an estimated $6,091-$10,000/month depending on utilization (20 vs. 40 hours/week), giving leadership clear cost visibility ahead of go-live.
- Unchanged designer workflow: The AWS Deadline Cloud submitter for 3ds Max and automated Egnyte network drive mounting preserved the team's existing submission and file-retrieval process, minimizing retraining.
- Documented, repeatable operations: A support playbook, including a real debugging walkthrough for a failed render output caused by a network drive credential change, equips YAGroup and ASCENDING support engineers to troubleshoot confidently.
- Risk-managed cutover: The rehosting approach allowed on-premise and cloud environments to run in parallel during transition, with a documented rollback plan reducing migration risk.
