Uploaded July 2026 | Updated November 2026, 5 hours ago
Are you looking to deploy Large Language Models (LLMs) locally without relying on expensive cloud APIs or sacrificing data privacy? In this video, Brett Kelly breaks down the best hardware/software options for running on-premise AI and local LLMs using enterprise-grade 45Drives infrastructure.
Whether you are hosting open-source AI models (like Llama 3, Mistral, or DeepSeek), fine-tuning custom models, or building secure RAG (Retrieval-Augmented Generation) pipelines for your organization, choosing the right server, GPU density, VRAM capacity, and high-speed storage is going to be critical.
READY TO BUILD YOUR ON-PREMISE AI INFRASTRUCTURE?
Talk to our team of storage and server engineers to architect a custom AI server tailored to your LLM workloads and budget: 45drives.com/contact
Explore 45Drives VM-Series Hardware: 45drives.com/products/storage-virtualizaton-server-proxinator-vm2
Learn about our on-premise AI solutions: 45drives.com/solutions/on-prem-ai
#OnPremiseAI #LocalLLM #45Drives #DataPrivacy #ArtificialIntelligence #SelfHostedAI #LinuxServer #EnterpriseStorage #OpenSourceAI #Llama3 #DeepSeek
Are you looking to deploy Large Language Models (LLMs) locally without relying on expensive cloud APIs or sacrificing data privacy? In this video, Brett Kelly breaks down the best hardware/software options for running on-premise AI and local LLMs using enterprise-grade 45Drives infrastructure.
Whether you are hosting open-source AI models (like Llama 3, Mistral, or DeepSeek), fine-tuning custom models, or building secure RAG (Retrieval-Augmented Generation) pipelines for your organization, choosing the right server, GPU density, VRAM capacity, and high-speed storage is going to be critical.
READY TO BUILD YOUR ON-PREMISE AI INFRASTRUCTURE?
Talk to our team of storage and server engineers to architect a custom AI server tailored to your LLM workloads and budget: 45drives.com/contact
Explore 45Drives VM-Series Hardware: 45drives.com/products/storage-virtualizaton-server-proxinator-vm2
Learn about our on-premise AI solutions: 45drives.com/solutions/on-prem-ai
#OnPremiseAI #LocalLLM #45Drives #DataPrivacy #ArtificialIntelligence #SelfHostedAI #LinuxServer #EnterpriseStorage #OpenSourceAI #Llama3 #DeepSeek

![How TIFF Actually Runs: Inside the Tech Behind a 450+ Film Festival on Whats Spinnin
What does it actually take to run a global film festival with 450+ films, thousands of screenings, and ZERO downtime?
In this episode of What’s Spinning?, Brett Kelly sits down with James King from TIFF to break down the real storage infrastructure, networking, and content management systems powering one of the world’s biggest film festivals. From DCP workflows and KDM key management to handling 4K films (250GB+ each) across 25+ screens, this is a deep dive into media storage, data pipelines, and high-performance server infrastructure.
Enterprise Data Storage & Virtual Compute Built on Open Platform& Open Source: https://www.45drives.com/products/hardware-storage-and-virtualization-servers/
The Premium Studio Line — Engineered for Filmmakers & Creators Who Demand More: https://45drivesstudio.com/
At 45Drives, we understand that enterprise customers value solutions that prioritize risk mitigation while being reliable, secure, performant, scalable, and high-capacity.
Follow Us for More:
LinkedIn: https://www.linkedin.com/company/45drives
Instagram: https://www.instagram.com/45drives
Facebook: https://www.facebook.com/45drives
X: https://www.x.com/45drives
[00:00] - Introduction and welcome to the Whats Spinning podcast.
[00:46] - Guest introduction: James King discusses his background, starting out building computers, and his 20-year journey as a cinema technician at TIFF.
[05:01] - The logistical challenges of screening 450+ films over 11 days and how IT infrastructure differs during festival season.
[08:20] - TIFFs storage management strategy: Centralized servers, library management systems, and the strict requirements of DCP (Digital Cinema Package) formats.
[10:44] - The critical need for redundancy to avoid downtime and ensure the show goes on.
[12:20] - Digital security, VLAN infrastructure, and managing thousands of KDM (Key Delivery Message) encryption keys to unlock films.
[15:19] - Storage capacities, bandwidth bottlenecks, and whether films are retained or purged after the festival.
[19:18] - The festival planning timeline, ingesting films, and the two-to-three week crunch.
[21:24] - Quality control, audio sync issues, and the rise of homebrew DCPs from first-time filmmakers.
[26:20] - Yearly infrastructure upgrades, maintenance windows, and the shift toward IP-based AV networking.
[29:29] - Taking over external venues using custom rolling racks equipped with their own networking gear.
[32:36] - The human element: How many IT and technical staff are required to keep the festival running smoothly.
[36:51] - The future of the industry: Shifting from physical hard drives to digital delivery and FTP portals.
[39:38] - Jamess thoughts on AI in the film and media exhibition space.
[43:16] - If he could change one thing: The need for standardized digital delivery methods across the industry.
[46:00] - Outro and wrap-up.
#torontofilmfestival #podcast #opensource #enterprisestorage #45drives How TIFF Actually Runs: Inside the Tech Behind a 450+ Film Festival on Whats Spinnin](https://i.ytimg.com/vi/YQjRzXaX9es/mqdefault.jpg)








