Protecting the Petabyte: Managing the New 'Blast Radius' in AI-Ready Infrastructure скачать в хорошем качестве

Protecting the Petabyte: Managing the New 'Blast Radius' in AI-Ready Infrastructure

Protecting the Petabyte: Managing the New “Blast Radius” in AI-Ready Infrastructure | Ben di Qual (Everpure) When your world moves from “big data” to petabyte reality, the rules change. AI doesn’t just scale your data… it scales the consequences of every decision you make about performance, locality, security, and recovery. In this episode of Azure Storage Talk, I’m joined by Ben di Qual (Everpure formally Pure Storage) for a conversation that’s not a product pitch—it’s a real look at what breaks first at scale, why “infinite cloud storage” is often misunderstood, and how leaders can shrink the blast radius in AI-ready infrastructure. We get into: Why “limitless capacity + performance” still hits real ceilings The locality gap (where compute lands vs where data can actually keep up) Why training ≠ inference (and why treating them the same gets expensive fast) What “AI-ready storage” should mean in practice (checkpointing, tiers, tradeoffs) Why incidents turn into board-level moments—and what restores trust afterward The practical “radius reducers”: boundaries, segmentation, immutability, recovery readiness 👇 QUESTION FOR YOU: What’s your biggest petabyte-era worry right now: performance, locality, security, or recovery? Chapters 00:00 Petabyte drives & the new “blast radius” 01:00 Welcome + why this is a “belief systems” conversation 02:00 The “infinite cloud storage” myth 03:00 Performance ceilings + the locality gap (GPU placement vs storage) 04:00 Lift-and-shift vs real cloud value 05:00 The missed step: refactoring + shared-nothing scale out 06:00 What “AI-ready storage” actually means 07:00 Checkpointing tiers + why old patterns are breaking 10:00 Marketing vs reality: training ≠ inference 11:00 The resource crunch (power, GPUs, NICs, lead times) 12:00 Pure + Azure: what’s exciting about the integration 13:00 After an incident: what leaders are really afraid of 16:00 Restoring trust: immutable snapshots + defense-in-depth 18:00 Balancing user autonomy with modern security risk 20:00 QLC + tradeoffs: doing more with less 21:00 Back to the petabyte: reducing blast radius at this scale 23:00 Power/footprint math + what happens when a server fails 24:00 What Ben wants you to check out next 25:00 Closing #AzureStorage #AIInfrastructure #CloudStorage #DataPlatforms #CyberResilience #DisasterRecovery #PureStorage #MLOps