Analyst(s): Camberley Bates
Publication Date: October 9, 2024

In Santa Clara, Tech Field Days hosted the first AI Data Infrastructure Days, bringing together Data Engineers and Data Storage Pros – what we can learn from the gathering.

What is Covered in this Article:

• How Object, File, and Block all play into generative AI
• Cloud Native and Multi-Cloud are part of the solutions
• Performance and scale are critical
• Security, governance, and protection take a key theme

The News: Tech Field Days hosted data engineering and data storage pros to review and discuss what is going on in data infrastructure and AI. With most of the technology press focused on GPUs, the data side has seen limited exposure, yet it is the core to training and inferencing. Vendors including Minio, Google, HPE, Infinidat, Solidigm, and Pure Storage (presenting in that order) provided their perspectives and capabilities to the delegates. This article provides a view into the takeaways.

AI Data Infrastructure Field Days – What We Can Take Away

Analyst Take: The last 12 months, since the launch of ChatGPT in the market, we have been in a frenzy of GPUs and is AI real or imagined. Tech companies and the stock market have all benefited from the rush to implement, yet for most AI is just getting started. A year in, and we are now seeing data – how it is managed, the compliance/governance, and the readiness of data sources for training become key topics within the scope of AI and generative AI.

Envisioning that there is no one path to effective data management and AI, Tech Field Day gathered a delegate team of data engineers (those entrenched in AI initiatives), system engineers (those working with the infrastructure) and data storage experts (those who been working with Block, File and Object) to hear and discuss the issues. In this first gathering of AI and Data Infrastructure, what can we take away in terms of issues and trends from the vendors and how they are approaching addressing the data problem for AI?

Some of the items are obvious; for others the issues are in the details. What did flow through the entire two days is data engineers and data storage infrastructure experts do not speak the same language. But, based on the conversation, this gap can be bridged with good common sense communication and open dialogs. So what were some of the themes we took away?

1. Object, File, Block – it all plays into Generative AI:
   • There appears to be two strains of thought. One, put all your data into one platform making management and speed of training/inferencing simpler. Or enable the integration of data sources maximizing the access to the data. With RAG (retrieval augmented generation), immediate access to new data will be a key requirement to keep the inference engine current, but that can add complexity. Still over the long haul, all protocols will likely need to be supported. How this plays out remains to be seen.
2. Cloud-Native Solutions:
   • The transition toward cloud-native architectures is a recurring theme. The vendors discussed how their solutions are designed to function seamlessly across on-premises and cloud environments, reflecting the blurred lines between the two. It is a nod to the reality of most enterprises that data resides scattered. It is another component on how the AI systems will be designed.
3. Integration of the AI Stack:
   • There was a focus on how the systems and storage supported the range of processes in the AI workloads. And a recognition that each process, i.e. LLM training, RAG, and Inference Engine, has different requirements of performance, capacity, and different tools used by the data analysts. This requires access to software and systems with integration to streamline the workflow including vector databases, GPUs, and data engineering tools for execution. Expect to see fully integrated systems as well as loosely coupled systems deployed.
4. Performance and Scalability:
   • With each presentation, we heard the ubiquitous issues around data growth and scale. The scale begins in the PB range, rapidly moves in EB, but can end up as an inference engine in the TB range. Treating AI as a single form of data infrastructure is fraught with errors as the vendors emphasized the importance of performance metrics, with innovations aimed at enhancing throughput and reducing latency. There were discussions around scaling storage solutions to meet massive data ingestion needs, particularly in sectors such as autonomous driving and threat detection. Then, there is the speed to keep the expensive GPUs operational plus the resulting read/write for real-time RAG. It is not one size fits all.
5. Security, Governance, and Protection:
   • Vendors were increasingly focused on security features, including encryption, data firewalls, and mechanisms for continuous data protection. This reflects a broader concern about cybersecurity in the data management landscape. There was also the understanding of protecting privacy and tracking where and how data was used to train. Overall, data management and concerns about data leakage and sovereignty were areas addressed.

What to Watch:

• Expect to see more collaboration between data engineers and the infrastructure teams as they tackle data problems and the overall architecture.
• Infrastructure teams will need to examine the trade offs of speed, ease of use, and flexibility as enterprises address the data issues.
• Security and governance along with privacy will remain top of mind as AI applications are rolled out.

Disclosure: The Futurum Group is a research and advisory firm that engages or has engaged in research, analysis, and advisory services with many technology companies, including those mentioned in this article. The author does not hold any equity positions with any company mentioned in this article.

Analysis and opinions expressed herein are specific to the analyst individually and data and other information that might have been provided for validation, not those of The Futurum Group as a whole.

Other insights from The Futurum Group:

Perspectives on the AI Data Pipeline with Solidigm – Six Five Webcast: Infrastructure Matters

How Does HPE Private Cloud AI Enable One-Click Deployment of GenAI Assistants?

Navigating AI’s Challenges and Opportunities – A Recap from The CIO Pulse Report

Author Information

Camberley Bates

Now retired, Camberley brought over 25 years of executive experience leading sales and marketing teams at Fortune 500 firms. Before joining The Futurum Group, she led the Evaluator Group, an information technology analyst firm as Managing Director.

Her career spanned all elements of sales and marketing including a 360-degree view of addressing challenges and delivering solutions was achieved from crossing the boundary of sales and channel engagement with large enterprise vendors and her own 100-person IT services firm.

Camberley provided Global 250 startups with go-to-market strategies, creating a new market category “MAID” as Vice President of Marketing at COPAN and led a worldwide marketing team including channels as a VP at VERITAS. At GE Access, a $2B distribution company, she served as VP of a new division and succeeded in growing the company from $14 to $500 million and built a successful 100-person IT services firm. Camberley began her career at IBM in sales and management.

She holds a Bachelor of Science in International Business from California State University – Long Beach and executive certificates from Wellesley and Wharton School of Business.