VMstore T7000 Series Icon

VMstore T7000 Series

On-prem VMstore platform.

Tintri Cloud Platform

Managed infrastructure powered by Tintri.

Tintri Cloud Engine

Container-driven VMstore platform.

All Die Hard Tintri News & Events Perspectives Products & Solutions Technical

2024: Inference Takes Flight. Now What?

It’s 2024- it’s here, it’s powerful, and it’s changing the application deployment landscape in a massive way. Highly trained inference applications will flood the market in 2024. These innovations require optimized platforms with complex distribution networks. Properly architected, order of magnitude gains in efficiency shall be realized. The long-term economic benefit to the bottom line is undeniable.

To deliver these business-critical applications, we will see a new evolution of application delivery platforms and progressive availability schemas. 2024 will witness a platform requirements explosion not witnessed since the late 1990s.

Brock Mowry, Tintri CTO & VP of Products notes:

 “In 2024, the platform demand for Generative AI applications will drive a new version of the ‘enterprise ready’ platform. Enterprise IT platforms will need to provide quick and powerful instances to support production, development, and playground environments. As a consequence, these instances will become the new challenge of workload sprawl within enterprises. The successful avoidance of such sprawl will dictate the business winners.”

It’s a fact, our world is written in the key of ‘do more with less.’ This is no less true in the world of AI. Inference applications are resource intensive. Paramount is the consideration of how businesses will compete when immense and high price systems are required to be successful.

Currently, AI processes and application management require highly specialized operators, complicated networking, complex co-compute needs, and complex storage features. To that end, we anticipate 2024 to be a year of incredible evolution in our current application and data delivery global initiatives.


Controlling Production Application Management Costs in 2024

We are staring into the infinitely expanding abyss of new and quite complex requirements. These innovative workflows will require new I/O paths, massively distributed compute distribution methodologies, scale up bandwidth requirements, and massive storage management demands. That’s before we get to the actual application deployment. Sometimes it stares back…Indeed.

Microservice architectures are by nature highly efficient when deployed on purpose architected platforms. Market beware, the converse is true and can be a progress killer on your journey. Scope early, build wisely.

Brock Mowry, Tintri CTO & VP of Products notes:

“Highly autonomous, evolved, Container platforms will play a crucial role in AI infrastructure in 2024, allowing for increased efficiency and flexibility, reducing total application TCO. However, the complexity of container infrastructure will pose a significant challenge for many enterprise IT teams. To bridge this knowledge gap, AI platform vendors will need to invest in user-friendly solutions and educational resources.”

“In 2024, the ability to manage and deliver resource granularity will be paramount. Monolithic, legacy based platforms will be replaced by ‘workload’ architected public and private AI platform solutions. Massive efficiencies vs. legacy are to be gained with employing this strategy.”


Virtual Machines will Drive the Flexibility and Segmentation of the AI Stack 

2024 will deliver the first new ‘killer’ generative AI and inference based critical applications at enterprise scale. We have spent a decade plus training neural networks to perform many complex tasks. Training modules are typically developed and executed outside of production application environments. Most often with non ‘production’ SLAs. Private AI in 2024 will unlock enterprise data and protect corporate IP from leaking.

This is one of the largest top three gating factors to broader adoption of AI within the enterprise.

Where will deployed apps live? VMs. VMs will be critical to accelerating the deployment phase of AI inference workloads. Once trained, we will see the “trained applications” reside on a VM topology whilst the “big iron” will find its home continually training the models.

Mark Walsh, Tintri VP, EMEA notes:

“In Europe we are seeing the adoption of VMs for production roll outs, sandbox, and CICD workloads. A VM inclusive topology allows our customers data scientists to seemly interact and deploy within traditional production application roll outs. We have seen massive improvements to our customers workflows. All signs point to a big ramp up in 2024.”


Evolving Security in 2024

While we welcome 2024, we also welcome new and improved AI and new AI based toolsets. As the availability of AI increases so will the number and more importantly, the increased capabilities they will have at hand. 

Brock Mowry, Tintri CTO & VP of Products notes:

“Bad actors are doing their R&D. 2024 will see new and undiscovered attack vectors leveraging the intelligence and power of AI. This will be the new adversary of blue teams in the enterprise.”

The age-old cat and mouse game has been greatly enhanced and we will have our collective hands full in 2024. It’s a brave new world!


The above content was originally published 1/4/2024 on VMblog.com.

This site is registered on wpml.org as a development site. Switch to a production site key to remove this banner.