Fixing My $10,000 Mistake - Whonnock 3 Deployment

The video chronicles a high‑stakes hardware deployment project nicknamed New Whonnock, documenting the drive to replace an underperforming storage server with a PCIe Gen 4 NVMe powerhouse capable of handling massive local storage and blazing fast data transfer. Early on, the team emphasizes the scale of the upgrade, aiming for 100 terabytes of storage on a 100 gigabit per second fiber network, and they frame the build as a race against time and budget constraints. The narrative unfolds with a detailed tour of Liqid’s Honeybadger carrier boards, AMD EPYC processors, and a modular storage architecture designed to maximize throughput while maintaining capacity. Throughout, the crew weighs tradeoffs between Gen 3 and Gen 4 SSDs, ultimately prioritizing capacity and endurance to support constant editing workloads, offsite backups, and rapid local transfers. The excitement comes with practical hurdles, like fragile cabling, chassis weight, and ensuring that the new system can feed multiple editors with stable latency and low contention. By the end of the episode, the team demonstrates tangible performance gains in real‑world editing tasks, confirming that the upgrade meets their expectations for latency, throughput, and reliability, even as they acknowledge the imperfections of large, complex systems. The second act of the video delves into connectivity and hardware integration, focusing on the ConnectX‑6 100 GbE NIC that provides dual 100 Gb/s ports and the promise of vastly higher per‑second data movement. The hosts explain how PCIe Gen 4’s high bandwidth is essential to exploiting the full potential of the Honeybadger carrier boards and M.2 drives, illustrating how the networking and storage subsystems must be tightly coupled to avoid bottlenecks. They also compare memory, parity schemes, and drive endurance, explaining their choice of Sabrent NVMe drives and the rationale for single parity protection given the workflow of frequent backups and offsite replication. The discussion shifts to assembly quirks and workflow rituals, including how the server will be deployed in a new rack and how the team will validate performance with live tests and real editors. The tone stays practical and humorous, with commentary about the “new new Whonnock” and the inevitable setbacks that come with ambitious upgrades. The segment culminates in a live throughput test, where latency and sustained transfer rates become the primary proof points for whether the upgrade has truly delivered on its promises. In the third act, real‑world testing takes center stage as the team runs multicam editing scenarios and probes whether the new storage and networking stack actually reduces latency and stabilizes playback. They move from synthetic benchmarks to tangible tasks, measuring throughput in gigabytes per second and translating those numbers into practical improvements for editors working with 8K material. The testers note a noticeable difference in responsiveness, with multicam playback and scrubbing behaving more predictably under load. The conversation turns to reliability concerns, as the team discusses how a tightly coupled network/storage path helps prevent cascading failures that previously torpedoed entire editing sessions. Moments of humor keep the mood light,jokes about cable fragility, the weight of the chassis, and the occasional misstep in the assembly process,while the core takeaway remains that the upgrade significantly improves editing workflow and reduces latency spikes. The hosts reflect on the broader implications for future deployments, acknowledging that while the project was expensive and complex, the results validate the investment for high‑demand media work. In the final stretch, the video credits the collaborators who made the upgrade possible, highlighting Liqid, Sabrent, Micron, FS.com, NVIDIA, and Infinite Cables, and calling out the 100 terabytes of fast storage as a milestone. They recap the performance gains, the improved editing latency, and the practical benefits of robust on‑site backups and offsite replication, tying the story back to the original goal of faster, more reliable production workflows. The hosts offer a light reminder that even with a triumphant rollout, ongoing maintenance and careful monitoring are essential to keeping such a sophisticated system running smoothly. The closing moments celebrate the team’s perseverance, the successful deployment of new Whonnock, and a nod to the journey ahead as the studio scales its high‑speed infrastructure for future projects.

Topics · data center · high performance computing · hardware engineering · networking · storage