KubeCon NA 2024: Service Discovery, Security, and AI

Though KubeCon North America 2024 has officially come to a close, the CNCF's flagship event has left us buzzing with residual excitement. After all, waving goodbye to a crowd of 9,200+ attendees is never easy—especially with Salt Lake City's snow-capped mountains towering impressively in the background.

However, it was our conversations with HAProxy booth visitors that truly stole the show. From the casual hello to the deeply technical dive, attendees eagerly shared their Kubernetes experiences and infrastructure challenges. Exploring these obstacles and corresponding tech trends was a tremendously rewarding experience. Our booth team lives for those "aha" moments where solutions meet user problems head-on and emerge triumphant.

Here's what we've learned from our four days alongside DevOps professionals, engineers, architects, and fellow K8s enthusiasts.

We've seen how Kubernetes (and open source software) constantly changes at HAProxy—as do the trending topics around it. Unsurprisingly, KubeCon North America 2024 demonstrated that such tech trends are as ephemeral as K8s pods. Our conversations with booth visitors repeatedly touched on some key topics:

AI and ML

We noticed an accelerated shift in interest this year towards AI/ML, which isn't shocking since 58% of organizations are actively experimenting with large language models (LLMs). And since today's experiments will likely be tomorrow's deployments, organizations are rightfully weighing load balancing options for AI/ML workloads. New ML training models and uses are emerging each day, and vendors like us have taken note. In fact, roughly 50% of KubeCon booths incorporated AI messaging in some way, shape, or form!

We fielded multiple related questions and observed an uptick in Kubernetes adoption specifically to support these applications. AI/ML is an exciting frontier we're eagerly exploring, which is why HAProxy One offers AI/API gateway support. Because our multi-cluster routing support is so capable, we're well positioned to support these bleeding-edge technologies as they mature.

Application security

Unsurprisingly, security remains a hot-button issue for the vast majority of organizations. KubeCon also demonstrated that while web application firewalls (WAFs) and bot management are critical, new and novel approaches to DDoS mitigation have garnered heavy interest. Customization and performance have become differentiators as users demand more from their security suites.

While K8s can do plenty on its own, security features such as HAProxy’s stick tables and HAProxy Enterprise's Global Profiling Engine (GPE) provide highly-effective supplemental protection against application-layer attacks. You can even deploy a stick table purely for bot-labeled requests to further reduce false positives.

The CAPTCHA module provides another layer of actionable protection, The module supports reCAPTCHA v2, reCAPTCHA v3, reCAPTCHA Enterprise, hCaptcha, Friendly Captcha (frCaptcha), and Turnstile. We chatted with many attendees who were eager to fill their existing security gaps with these features.

Service discovery

Service discovery has always been integral to uncovering active Kubernetes services and pushing configuration changes on the fly. While the concept isn't new, software vendors offer slightly unique flavors of service discovery with varied levels of automation and performance.

We chatted with many booth visitors who enthusiastically discussed their discoverability needs while expressing a desire for improved scalability for their K8s services. Our team showcased our high-performance service discovery (introduced in HAProxy Fusion 1.2 and improved in HAProxy Fusion 1.3) to great fanfare in Salt Lake City—fueling our fire to further improve the feature.

With the power to automatically generate over 100,000 lines of HAProxy configuration in seconds, HAProxy Fusion service discovery is a perfect match for large-scale deployments. We're thrilled to continually improve upon this key Kubernetes capability and showcase our progress at the next KubeCon. Don't miss it!

Answering your K8s questions

KubeCon was a massive convergence of ideas and curiosities, all flowing like a firehose without an off valve. We quickly learned that vendor lock-in remains a primary concern for plenty of organizations, and simplicity is a guiding principle for many grappling with Kubernetes complexities. Attendees also perked up for some exciting new development avenues, such as egress gateways and TLS-based SNI allowlisting using access control lists (which one of our customers is already doing successfully).

Naturally, attention soon turned to us and our development roadmap. Visitors kept us occupied with numerous questions about our platform and vision behind the evolving HAProxy One platform. Here are responses to some common questions our booth team received:

Does HAProxy offer Ingress functionality?

Yes! HAProxy's comprehensive Kubernetes solution includes Kubernetes Ingress support for organizations requiring simple setup, low resource use, high performance, and cost efficiency.

Ingress control exists alongside our unique approach to intelligent external load balancing, multi-cluster routing, and blue-green deployments. Deploy HAProxy Enterprise Kubernetes Ingress Controller independently or together with other components in our K8s solution, according to your load balancing needs.

These features are fulfilled by different products within HAProxy One, so we'd love to chat and determine what best fits your needs.

With Ingress' development "frozen," is Gateway API the way forward?

We're continually evaluating Gateway API support in HAProxy One and plan to bring Gateway API to HAProxy Fusion Control Plane. We want to support as many customer use cases (and preferences) as possible. We also anticipate that organizations will increasingly migrate away from Ingress to an alternative solution.

Our Kubernetes solution enables external load balancing and multi-cluster routing without the added complexities of Gateway API. HAProxy enables you to route traffic directly from your external HAProxy Enterprise nodes to your Kubernetes pods without having to use Ingress or Gateway API at all.

There's no management overhead with vendor-specific policies, nor a need to install additional custom resource definitions (CRDs) unless that helps your use case. Using HAProxy Fusion to automate direct-to-pod load balancing also eliminates a network hop normally associated with querying Gateway API. This reduces latency for massive-scale K8s applications while removing a potential point of failure.

We're truly excited to see how HAProxy Enterprise and HAProxy Fusion service discovery can help our customers' applications perform better at massive scale. This approach is future-proof and a great next step for users should Ingress reach end-of-life.

Thanks for engaging with us

As always, your questions excited us, challenged us, and have even inspired us to redefine what's possible with Kubernetes and HAProxy One. It was great catching up with fresh and familiar faces alike. We never get tired of taking visitors through live demos or deep whiteboard sessions—blending plenty of technical knowledge with a little artistic flair.

The HAProxy community also deserves a gigantic shoutout for its willingness to share valuable feedback. KubeCon left us (happily!) drowning in a tidal wave of G2 reviews, which help us improve HAProxy One and prioritize popular feature requests. Please, keep those opinions coming and your voices loud!

Come see us next year!

KubeCon North America 2024 blew us away. Our conversations have helped us better understand the evolving needs of the K8s community and better position ourselves as a leader in container networking.

It's now time to flip the page to KubeCon Europe 2025 and KubeCon North America 2025. We can't wait to unveil some exciting new developments and see how the Kubernetes landscape changes. We'll also be at AWS re:Invent 2024 in Las Vegas, from December 2nd to December 6th. Come see us at Booth 571!

Last but not least, we're thrilled to kick off HAProxyConf 2025 next summer in San Francisco. HAProxyConf celebrates the thriving community that's helped make HAProxy One the world's fastest application delivery and security platform. Over 2+ days, expert speakers will share best practices and real-world use cases that highlight HAProxy's next-gen approach to high-performance application delivery and security. Check out the HAProxyConf official website to learn more, stay updated, and answer our call for papers!

Want to learn more about HAProxy and Kubernetes?

To dive a little deeper into our Kubernetes solutions and story, check out these helpful resources:

Our products and HAProxy One—the world’s fastest application delivery and security platform—are always evolving. Stay tuned for important updates and development milestones! Thank you for another fantastic KubeCon.

Subscribe to our blog. Get the latest release updates, tutorials, and deep-dives from HAProxy experts.