Name: AI Model Inferencing and Deployment Options
Start: 2025-11-13T18:00:00Z
End: 2025-11-13T18:01:01.000Z
Location: BrightTALK

The Enterprise Storage channel has the most up-to-date, relevant content for storage and infrastructure professionals. As data centers evolve with big data, cloud computing and virtualization, organizations are going to need to know how to make their storage more efficient. Join this channel to find out how you can use the most current technology to satisfy your business and storage needs.

In today’s evolving IT landscape, selecting the right storage architecture is critical for optimal performance, scalability, data governance, and cost-efficiency. Furthermore, AI workloads have uniquely influenced how we meet these demands from our storage infrastructure. This webinar provides a technical deep dive into three fundamental storage deployment models – on-premises, cloud, and hybrid – examining their architectures and operational trade-offs through the lens of two key concepts: indirection (accessing data through mapping layers that provide flexibility and abstraction) and redirection (rerouting data requests to enable failover, load balancing, and optimized performance). 

We are going to take some of the key stages of AI lifecycle development as sample use-cases (such as data ingestion, preparation, training, inferencing, and retrieval) and compare how each storage model can serve these use-cases across varying access patterns, data volumes, and performance requirements.

Attendees will gain a practical framework for aligning AI workloads with the most suitable storage architecture, balancing cost, scalability, and latency. Whether you are building AI infrastructure from scratch or optimizing existing deployments, this session will help you make informed decisions for AI-ready storage.

Learning Objectives and Key Takeaways:

• Introduction to the 3 different types of storage deployment models – on-prem, cloud, and hybrid
• Trade-offs for each deployment model
• Importance of indirection and redirection
• Understand how AI-specific data types and access patterns (e.g., embeddings, checkpointing) influence storage performance and design
• Evaluate trade-offs in latency, scalability, security, and cost when choosing storage for different stages of the AI pipeline
• Gain a decision-making framework for selecting the right storage model based on workload characteristics and infrastructure goals

AI Meets Storage: Comparing On-Prem, Cloud, and Hybrid Architectures Across the AI Lifecycle

As AI workloads continue to scale, the physical infrastructure supporting them must evolve to deliver the performance, reliability, and efficiency required by modern data-intensive applications. This webinar will focus on the interconnect foundations of AI infrastructure, including storage. There will be an emphasis on the form factors, connectors, cables, and transceivers standards that enable scalable and interoperable systems.

We’ll explore how SNIA’s SFF Technical Work Group is contributing to the development of physical layer standards that support high-performance interconnects and storage devices used in AI environments. Topics will include the importance of transceiver innovation and the need to evolve; the role that form factors like EDSFF play; and how innovative cabling and connector designs support the adoption of high-speed signaling technologies like PCIe 8.0 and beyond.

This session is ideal for architects, engineers, and technologists who want to understand how physical infrastructure standards are enabling the next generation of AI systems.

Key Concepts:
• Transceiver development for out-of-box signaling 
• Connectors and cables for high-speed inbox signaling
• Copper vs. optical and the challenges of both
• SSD form factor innovations and their impact on density and thermal design

Accelerating AI Infrastructure: The Role of 400G and PCIe 8.0 in Next-Gen Interconnects

Storage security changes and adaptations are a fact of life to deal with the ever-changing threat and regulatory landscapes. A black swan event, a new standard, or a regulation often serves as a catalyst for organizations to review their controls and practices. The recent publication of NIST Special Publication 800-88 Rev. 2, Guidelines for Media Sanitization, is drawing attention to storage security, but it is not the only development worth noting. SNIA, Open Compute Project (OCP), Trusted Computing Group (TCG), and IEEE are developing or have developed standards and specifications that could be important going forward. One central theme underlying several of the activities is in storage sanitization (i.e., controlled eradication of data).

This session brings together a unique panel of experts who have served as editors of some of the most important storage sanitization standards. These experts also have broad knowledge of storage security. As a panel, they will provide insight into these new developments as well as observations on what organizations are experiencing. A moderator will engage the panel with thought-provoking questions, but participants are also encouraged to bring their own questions.

Join us to understand: 
• Changes to storage sanitization and what that means to organizations
• Primary drivers for storage security
• Difference between Storage Sanitization and Data Sanitization
• Interplay and interdependencies for storage security standards/specifications

The Storage Security Shake-Up: Adapt Now or Get Left Behind

Magnetic RAM, Resistive RAM, Ferroelectric RAM, and other new memory technologies have been pushed to the forefront by issues with Moore’s Law scaling, but the ancillary benefits they bring to computing will provide significantly more important longer-term benefits.  In this session, analysts Jim Handy and Tom Coughlin will explain how these new nonvolatile technologies will help accelerate and cost-reduce inference through compute-in-memory engines while standard computing will also see lower costs and higher performance by bringing inexpensive persistence closer to, and even within the processor chip itself.  The session will also address ways that new frameworks like CXL will help provide persistence-friendly structures at all levels of the computing hierarchy.

How New Memories Will Accelerate Both Inference and General-Purpose Compute

Smarter data management is the key to unlocking the full potential of next-generation data infrastructure for AI. This webinar explores how the SNIA Cloud Data Management Interface (CDMI™) standard enables government labs, HPC centers, and organizations leveraging the power of AI/ML to reimagine and innovate their infrastructure to meet next-generation data management demands. CDMI is a mature ISO standard (ISO/IEC: 17826:2022) with over 15 years of development and implementation, and is purpose-built for cloud data management. CDMI's extensive capabilities enhance key aspects of AI operations by transforming data management challenges into streamlined, intelligent workflows. 

CDMI 3.0 (in development now) will further extend the standard with:
• Automated resource discovery that accelerates AI workload efficiency
• Portable data movement that enables preservation of metadata fidelity during migrations, including graph relationships and knowledge graph
• MCP-based data management that enhances AI accuracy while improving access speed

Building on a proven, standards-based foundation, CDMI enables true machine-readable data management. CDMI 3.0 will further improve how teams discover resources, orchestrate AI agents, and protect data across complex cloud environments while establishing the infrastructure for automated, self-describing data ecosystems with full auditability that drive future innovation. 

Read the SNIA CDMI 3.0 white paper https://www.snia.org/educational-library/introduction-cloud-data-management-interface-cdmitm-30-2025

Intelligent Data Management: Shaping the Future for AI Workloads

Join us for an insightful webinar and demonstration as we explore the foundational process of training machine learning models, with a special focus on computer vision applications. This session will guide participants through the essential stages of model development—from data preparation to algorithm selection—highlighting how these steps influence performance and accuracy.
Through practical demonstrations and conceptual discussions, attendees will gain a deeper understanding of how vision models are built and validated, and how these techniques can be extended to broader predictive analytics. We’ll also examine common challenges and share strategies to overcome them, emphasizing the value of simplicity, modular design, and thoughtful evaluation.

Key Takeaways:
• Core steps in the model training lifecycle
• Why data quality is critical to success
• Understanding inferencing and metrics for model validation
• Shifting from engineering mindset to modeling intuition
• Avoiding common pitfalls in training workflows
• Embracing parsimony and modularity in design
• Using checkpoints to monitor and refine progress

From Data to Decisions: Understanding How AI Models Learn

Storage networking consumers are fully aware of the classic benefits of Fibre Channel as a dedicated storage networking solution that provides reliable and lossless data transfer as well as high availability and redundancy.  What is less known is that Fibre Channel supports several Upper Layer Protocols (ULPs) including SCSI, FICON, NVMe, Avionics and RDMA. In this webinar, we explore the FC-4 layer of the Fibre Channel architecture that provides the services used by the various ULPs deployed in Fibre Channel products. Join us for an exploration into:

• The history of Fibre Channel’s versatility 
• Fibre Channel architectural layers
• Overview of the current FC-4 transports: FCP, FC-SB, FC-NVMe, FC-RDMA

Attendees of the webinar will become well versed in the scope and capabilities of Fibre Channel and develop a well-rounded understanding of the technology.

Versatility of Fibre Channel: Multi-Protocol Magic

In today’s data-driven world, ensuring the resilience of your storage systems is critical for long-term success. This vendor-neutral webinar, hosted by the STA, will bring together industry experts to explore proven methodologies for testing the robustness of Serial Attached SCSI (SAS) and other data storage infrastructures. Craig Foster (Teledyne LeCroy) and Paul Coddington (Amphenol) will discuss practical approaches and the latest standards that enable secure, reliable, and high-performance storage systems. Rick Kutcipal (Broadcom) will moderate the Q&A and offer other perspectives. Join us to gain insights on ensuring your storage solutions meet the demands of modern workloads and prevent costly data loss or downtime.

Key Takeaways:

- Understand the key factors impacting the robustness of Serial Attached SCSI (SAS) and other data storage systems.
- Learn about the latest testing methodologies and standards in SAS and storage infrastructures.
- Discover how vendor-neutral best practices can enhance the security and reliability of SAS-based storage.

Reduce Your Risk of Data Loss! "Critical Testing Insights for Developers"

This webinar will provide a foundational understanding of Artificial Intelligence (AI) and Machine Learning (ML). We will explore the basic concepts, history, and applications of AI/ML, setting the stage for more advanced topics in subsequent webinars. Attendees will gain insights into how AI/ML technologies are transforming industries and everyday life and have the opportunity to see one of the earliest AI experiments: Training a neural network to recognize written letters.
 
Key Concepts:
• Definition and History of AI and ML
• Differences Between AI, ML, and Deep Learning
• Real-World Applications of AI/ML
• Basic Terminology and Concepts

Introduction to AI and Machine Learning

In this webinar, we’ll introduce the PCI Express (PCIe) interface which is an essential component of AI systems due to its ability to provide high bandwidth and low latency over short distances. We’ll introduce the basics, then discuss the evolution, requirements and benefits. We’ll cover features in PCIe Gen6 including bandwidth, AER (Advanced Error Reporting), DPC (Downstream Port Containment), different modes and signaling techniques. We will also cover PCIe switching and its applications for AI. Join us for an insightful discussion on:
- What is PCIe and why it matters? 
- Discovery/Enumeration and Hot-plugging 
- NTB (Non-Transparent Bridging) applications and uses
- Shared credit/received buffer
- Selective NAK vs. Standard NAK for Flits replay
- PCIe Gen6 Enhancements
- PCIe Switch and applications for AI

Everything You Wanted to Know About PCIe But Were Too Proud to Ask

The rapid growth of AI is driving unprecedented demand for high-bandwidth, high-capacity memory solutions. Compute Express Link (CXL) is emerging as a key enabler in 2025, unlocking new levels of memory scalability and efficiency. Join experts as they discuss how CXL-based solutions are addressing memory bottlenecks and capacity constraints in modern data centers.
 
Attendees will:
Learn about real-world applications and customer platforms using CXL
Gain insights into benchmarks for CXL performance and latency
Understand how OEM customers are leveraging the CXL solution

Accelerating AI with Real-World CXL Platforms

During SNIA's Cloud Object Storage (COS) multi-vendor compatibility plugfests, it became apparent to the community that a lack of commonly accepted open-source COS test tools is one of the more obvious pain points for the industry. The consensus from the plugfest participants is that we need to collaborate on test tools because often tools are proprietary and biased, and given the number of tools, it is nearly impossible to test them all. Participants found that working together, using other people's test code and results, was highly beneficial, providing broader coverage vs. testing solely with their own tools.

With this ongoing COS Plugfest effort, SNIA member companies aim to create and publish COS interoperability test software suites that help automate protocol compatibility testing for popular industry cloud object storage APIs, wire protocols, and services. If you’re curious about cross-vendor S3 compatibility, join us to learn about the work within our community. 

In this webinar, experts engaged in proprietary and/or open-source software development will discuss:
· The real-world need for COS interoperability test tools
· Examples of compatibility gaps
· Highlight open-source test suites (e.g., Ceph, Versity and others) 
· Progress to date from ongoing testing
· How to get involved

SNIA Cloud Object Storage Test Tools (Open-Source Industry Interoperability Project)

Agentic AI is an emergent and disruptive paradigm beyond generative AI that carries great promise and great risk. With generative AI, a large language model (LLM) generates content as the result of a prompt. With agentic AI, the language model now generates actions to execute on behalf of a user using tools at its disposal. It has agency to act on your behalf and even run autonomously.  But how can this be done safely and effectively for enterprises?
 
This webinar will introduce Agentic AI and cover:
• What is agentic AI?
• What makes it different from generative AI?
• Adoption challenges and what to consider when evaluating solutions for enterprise use
• Sample use case
 
Join us to understand both the risks and benefits of agentic AI for enterprises use cases.

Agentic AI – Use Cases, Benefits, Risks

Explore cutting-edge innovations in solid-state drive (SSD) power efficiency and liquid cooling, designed to mitigate AI workload bottlenecks and reduce Total Cost of Ownership (TCO) in data centers. By focusing on optimizing SSD performance per watt, we can significantly enhance data center sustainability, operational efficiency, and economic viability.
Key topics include:
1. NVMe® Power Tuning Techniques: Maximizing performance per watt through advanced power management strategies.
2. SSD Controller Architecture Breakthroughs: Leveraging hardware offloads and computational storage innovations to improve energy efficiency.
3. Liquid-Cooled SSD Technologies: Pioneering advancements in liquid cooling that enable significant reductions in power consumption and facilitate the transition to fan-less server designs.
By examining the latest trends in immersion and liquid cooling technologies, this presentation will demonstrate their critical role in driving data center performance, sustainability, and cost-effectiveness forward. Join us to discover how these advancements can help overcome challenges in high-density AI environments.

Unlocking Sustainable Data Centers: Optimizing SSD Power Efficiency and Liquid Cooling for AI Workloads

SNIA's Cloud Object Storage community organized and hosted the industry first open multi-vendor compatibility testing (Plugfest) at SDC’24, in Santa Clara, CA. Most of the participants focused on their S3 server and client implementations. Community driven testing revealed significant areas to collaborate, including ambiguities among protocol options, access control mechanisms, missing or incorrect response headers, unsupported API calls and unexpected behavior (as well as sharing best practices and fixes). The success of this event led to community demand for more SNIA Cloud Object Storage Plugfests.

In this webinar, SNIA Cloud Object Storage Plugfest participants will discuss real-world issues hampering interoperability among various S3 implementations, from both server and client perspectives, covering:

• Goals and collaborative efforts of this multi-vendor community
• S3 compatibility challenges and why they happen
• This Community’s effort to identify and fix issues at SNIA Cloud Object Storage Plugfests
• The value of multi-vendor compatibility testing

Community Driven S3 Compatibility Testing

Join Jim Handy of Objective Analysis, Mahesh Natu of the CXL Consortium, and Torry Steed of SMART Modular Technologies as they delve into Compute Express Link® (CXL), a groundbreaking technology that expands server memory beyond traditional limits and boosts bandwidth. CXL enables seamless memory sharing, reduces costs by optimizing resource utilization, and supports different memory types. Discover how CXL is transforming computing systems with its economic and performance benefits, and explore its future impact across various market segments.

Unlocking CXL's Potential: Revolutionizing Server Memory and Performance

In today's world of AI/ML, high-performance computing, and data centers, RDMA (Remote Direct Memory Access) is essential due to its ability to provide high-speed, low-latency data transfer. This technology enhances scalability, reduces CPU overhead, and ensures efficient handling of massive data volumes, making it crucial for modern computing environments. But how does it actually work?

In this session, we'll define RDMA, explore its historical development, and its significant benefits. In addition, we’ll provide an in-depth look at RDMA over Converged Ethernet (RoCE) and InfiniBand (IB), cover the fundamental objects, Verbs API, and wire protocol. 

Key Takeaways:
1. Insights into the historical development and evolution of RDMA technology.
2. Comprehensive understanding of RDMA and its role in modern computing.
3. Detailed exploration of RDMA implementations
4. Explanation of fundamental RDMA components, such as the Verbs API and wire protocol (e.g., IB and RocE)

Join us for an insightful SNIA Data, Storage & Networking webinar on RDMA.

Everything You Wanted to Know About RDMA But Were Too Proud to Ask

This webinar will explain the process of model inferencing and deployment options. We will discuss how trained models are used to make predictions and the considerations for deploying models. Attendees will understand the challenges and strategies for optimizing model performance, covering:
 
• What is model inferencing  
• Model inferencing process 
• How inferencing deployment differs in Gen AI and visual inspection
         - Inferencing deployment software options 
          - Inferencing hardware considerations (e.g., on the edge, on-prem, etc.)
• Deploying AI models for production best practices and lessons learned 
• Strategies for maintaining model performance post-deployment

AI Model Inferencing and Deployment Options

The storage community on BrightTALK is made up of thousands of storage and IT professionals. Find relevant webinars and videos on storage architecture, cloud storage, storage virtualization and more presented by recognized thought leaders. Join the conversation by participating in live webinars and round table discussions.

Storage

As an IT professional, many of the problems you face are multifaceted, complex and don’t lend themselves to simple solutions. The information technology community features useful and free information technology resources. Join to browse thousands of videos and webinars on ITIL best practices, IT security strategy and more presented by leading CTOs, CIOs and other technology experts.

AI Model Inferencing and Deployment Options

Presented by

About this talk

Storage