Skip to main content

[@ServeTheHomeVideo] Modernizing Enterprise Applications for AI - Open Storage Summit 2025 Session 5

· 7 min read
[@ServeTheHomeVideo] Summarizer
Video Bot

@ServeTheHomeVideo - "Modernizing Enterprise Applications for AI - Open Storage Summit 2025 Session 5"

Link: https://youtu.be/-mCongb8zbc

Short Summary​

Okay, here's a summary of the transcript:

Number One Takeaway: Enterprises modernizing applications for AI need to consider end-to-end, scalable, and power-efficient infrastructure solutions to simplify deployment, manage complexity, and optimize performance as they transition from AI modeling to real-time AI applications at scale.

Executive Summary: The shift to AI-driven applications requires enterprises to modernize their legacy systems with open, cloud-native architectures built on platforms like Postgress, high-performance block storage, and Kubernetes. Super Micro, EDB, AMD, and Lightbits offer integrated solutions that simplify AI deployment, reduce TCO, and provide the scalability and performance needed to support demanding AI workloads, emphasizing the importance of open standards, ease of use, and a strategic, patient approach to AI adoption.

Key Quotes​

Here are five quotes extracted from the video transcript that I found particularly insightful:

  1. Rob Stretch: "As AI reshapes the enterprise technology landscape, application modernization is no longer a choice. It's a competitive necessity."
  2. Jun Shia Xiao: "In the past few years, a couple dozen companies have led the charge in developing and training AI. And now we are seeing a bigger shift that enterprises are adopting AI at a scale. And so this transition from the AI modeling to inference and uh real time applications requires a new approach to computer infrastructure."
  3. Shiva Guru Murthy: "So AMD has become is become a building block of all the stuff that Junjier has talked about uh previously. Uh we have a broad broadest portfolio of AI related products in the industry."
  4. Sige Bulov: "They want to have as much as many IO operations in as little as possible uh footprint in their data centers."
  5. Simon Lightstone: "So at EDB we provide both enterprise editions of Postgress with many other capabilities and we also pro provide paid support for companies and we have a new Postgress data and AI platform which has a lot of exciting capabilities and that make databases and AI work all together and let you build AI apps very easily."

Detailed Summary​

Okay, here's a detailed summary of the YouTube video transcript, focusing on the key information and arguments presented:

  • Introduction & Context:

    • Super Micro has been providing computing resources to innovators for over 30 years.
    • The video is part of Super Micro's 2025 Open Storage Summit.
    • The focus is on modernizing enterprise applications for AI.

  • The Core Argument: Application Modernization for AI

    • Application modernization is now a "competitive necessity" due to the demands of AI.
    • Traditional monolithic systems and legacy databases are struggling.
    • Enterprises are shifting towards open, cloud-native architectures, with platforms like PostgreSQL at the center.

  • Key Enablers for AI-Driven Application Modernization:

    • PostgreSQL:
      • Extensible, cost-effective, open ecosystem support.
      • Enterprises need to rethink performance at all layers of the stack to utilize Postgres' full potential.
    • High-Performance Block Storage:
      • Critical for low-latency, high-throughput access for transaction-heavy PostgreSQL workloads.
      • Essential as AI inference and OLTP processing converge.
    • Kubernetes:
      • A "foundational control plane" for deploying, scaling, and managing modern OLTP environments.
      • When combined with AI-enhanced microservices, enables cloud-native, data-driven, real-time insight applications.

  • Super Micro's Role (Jun Shia Xiao):

    • Super Micro helps businesses move from AI experimentation to full-scale deployment.
    • Addresses pain points in AI infrastructure deployment:
      • Complexity in design, deployment, and management.
      • High energy consumption and cooling challenges.
      • Networking complexity.
    • Super Micro's "Rack Scale" solution:
      • Consultation to determine platforms and components.
      • Solution design based on workloads and power budget.
      • Rack assembly and validation before shipment.
      • Ongoing on-site services.
    • Rack scale market expected to grow significantly.
    • Super Micro offers total solutions, from systems to rack clusters and data center building solutions.
    • Strategic partnerships (EDB, Lightbits, AMD) enable comprehensive technologies.

  • AMD's Role (Shiva Guru Murthy):

    • AMD is a "building block" for AI systems.
    • Offers a broad portfolio of AI-related products.
    • AMD EPYC CPUs are central to many AI systems, both as host nodes for GPUs and for inference.
    • AMD Instinct GPUs for training and inference.
    • AMD Pensando DPUs for networking and data transfer.
    • Supports open standards like Ultra Accelerator Link.
    • AMD Versal FPGAs for efficient inference.
    • AMD has significant market share (40%) and is trusted by large companies.
    • Strong performance in PostgreSQL workloads (OLTP and query-based analytics).
    • Cost savings: achieving better performance with fewer CPUs per server.

  • Lightbits' Role (Sige Bulov):

    • Lightbits is a software-defined storage solution.
    • Focuses on running on small footprints and providing millions of IOPS.
    • Three pillars:
      • NVMe over TCP: Protocol invented by Lightbits.
      • Intelligent Flash Management: Enhances SSD lifecycle and performance.
      • Clustering: Provides resiliency with configurable replicas.
    • Works with various Kubernetes flavors (Red Hat OpenShift, vanilla Kubernetes, etc.).
    • Also offers drivers for OpenStack, VMware (vSphere/VCF), and bare metal deployments.
    • Performance: Demonstrates scalability with EDB CNPG on Super Micro X14 platform with AMD CPUs, achieving high transaction rates.

  • EDB's Role (Simon Lightstone):

    • EDB is the number one contributor to PostgreSQL.
    • Offers enterprise editions of PostgreSQL and paid support.
    • Provides a PostgreSQL data and AI platform.
    • Powers mission-critical systems.
    • Partnership with Super Micro addresses:
      • Mission-critical performance: Curated configuration for optimal performance.
      • Extreme high availability: Simple solution for data center resilience.
      • Simplicity: Reduces need for specialized training.
      • Integration with AI: Makes AI easy to deploy and ensures sovereignty.
    • Sovereign Data & AI Platform:
      • Integrated stack of hardware and software.
      • Simplified deployment of highly available PostgreSQL clusters and AI chatbots.
      • Full lifecycle support (security updates, major version updates).
    • Offers six times the ROI on database investment.
    • Claims 150x faster that MongoDB, 4x Faster than MySQL

  • Adopting AI at Scale (Jun Shia):

    • AI is driving demand for optimized hardware (compute, networking, storage).
    • Enterprises want end-to-end deployable AI infrastructure (on-premise, hybrid, cloud).
    • Power and cooling efficiency are critical.
    • Super Micro's capabilities:
      • In-house engineering and manufacturing.
      • Rack-scale plug-and-play solutions (one-stop-shop).
      • Leadership in liquid cooling technology.

  • Energy Efficiency (Shiva):

    • Power and data center space are major constraints.
    • AMD focuses on energy-efficient platforms.
    • Better performance means customers achieve more with less footprint.
    • Consolidation with AMD servers: Up to 7:1 consolidation possible.
    • Significant reductions in power consumption and overall TCO.

  • Ease of Use (Sige):

    • Ease of use is crucial for Lightbits customers.
    • Easy installation on various Linux distributions.
    • Consistent addition of servers and updates.
    • Software-defined storage provides flexibility.

  • Personas and Simplification (Simon):

    • Many specialized roles are involved in AI (developers, data scientists, hardware specialists).
    • AI systems (especially due to energy consumption) can be expensive, so optimization is important.
    • AI developers should focus on creating results, not on patching and low-level tasks.
    • EDB Data and AI Platform and the partnership with Super Micro help to focus on higher-level tasks and provide a modern stack.

  • AI Factories and TCO (Group Discussion):

    • Right-sizing infrastructure is key to lowering TCO.
    • Lightbits focuses on high performance and minimal footprint.
    • EDB emphasizes speed to production.
    • Super Micro provides a flexible portfolio.

  • Key Takeaways (Final Statements):

    • Sige (Lightbits): Data is most important; ensure a capable, scalable, modern storage subsystem.
    • Simon (EDB): Build on open standards, and begin with the end in mind (production).
    • Shiva (AMD): Be patient; AI is a journey; start with pilot projects.
    • Jun Shia (Super Micro): AI has changed how businesses approach infrastructure; Super Micro simplifies deployment and enables expansion.