Red Hat Summit 2025

This year I was incredibly fortunate to visit historic Boston for Red Hat Summit 2025. If you haven't attended Red Hat Summit before, it's Red Hat's flagship event of the year. It's where Red Hat customers and partners can hear about the latest innovation, see demonstrations of new capabilities in action, and join communities like OpenShift Commons and AnsibleFest to hear how others are using open source technology across the hybrid cloud.

This year Red Hat Summit was held at the Boston Convention and Exhibition Centre (BCEC). Fun fact - BCEC is the largest building in New England, and is as long as the Empire State Building is high. Image from Rafael Vinoly Architects:

Announcements

Red Hat Summit kicked off on the Tuesday morning with a number of announcements. You can get a glimpse here of what it was like walking into the amphitheatre.

And a pic inside:

llm-d

Artificial intelligence (AI) was a key focus at Red Hat Summit this year, and my favourite announcement was the creation of the open source llm-d project. Brian Stevens, previously Red Hat CTO and now back at Red Hat following the Neural Magic acquisition, announced this on-stage. I snapped a pic here:

So what is llm-d? One of the challenges with generative AI inference is that everything happens on the same accelerator (like a GPU), on the same server. This makes it difficult to scale resources needed to support AI-enabled applications. llm-d solves these challenges by introducing a model for distributed inference; splitting the inference process into two distinct phases, prefill and decode, and running each in separate pods on Kubernetes (OpenShift). llm-d also includes distributed inference optimisations, like KV-cache aware routing and disaggregated serving. You can find a video here showing llm-d in action:

You can also read a blog here about why llm-d is a gamechanger for AI inference from one of Red Hat's AI experts in Singapore, Li Ming Tsai. Another fun fact - Li Ming and I both joined Red Hat in the same month back in 2016, and were both part of the professional services organisation working on the same product, Red Hat CloudForms.

Red Hat AI Inference Server

Another exciting AI announcement was the release of Red Hat AI Inference Server. This is the enterprise-grade, Red Hat-supported product derived from the vLLM open source project, which is now effectively the standard runtime now for AI inference. You might have already seen some of the innovation Red Hat and Meta are building with vLLM already, like day zero support for Llama 4.

I know many of you reading this might be security professionals, and read that as "zero day". But "day zero" is very different - it means that when a new LLM drops from Meta, or other organisations, it already has engineered support within vLLM (Red Hat AI inference server). This means that you don't need to wait for the runtime to catch up with the model - it supports it from "day zero". How great is that!

vLLM is a really interesting technology. It was born out of a research paper that included researchers from UC Berkeley, UC San Diego and Stanford University, titled Efficient memory management for Large Language Model serving with PagedAttention. The paper demonstrates how existing LLM serving runtimes are limited in how they manage something called "KV cache", a popular technique to speed up LLM inference. KV cache suffers from some of the same problems that operating systems do when they run out of working memory, and start "paging" - swapping memory pages to the disk, and back to working memory. You can see how KV caching works in this diagram from huggingface:

The vLLM paper introduces a new algorithm inspired by paging mechanisms of operating systems to manage memory cache, PagedAttention, that results in 2-4x throughput improvement.

[NB: I mentioned llm-d above. One of the core llm-d capabilities is distributing KV cache lookups across multiple nodes]

If you want to read more about Red Hat AI Inference Server check out the article here.

Post-quantum cryptography

Another one of my favourites was from Ashesh Badani, announcing support for post-quantum cryptography (PQC) in Red Hat Enterprise Linux (RHEL). RHEL now includes a system-wide post-quantum cryptographic profile, enabling you to enable support for some of the NIST-approved PQC ciphers currently available.

Why this is important is because the timeline for a cryptographically-relevant quantum computer is getting scarily close. By 2030 it's expected that a cryprographically-relevant quantum computer will be able to break classical encryption, including factoring RSA and retrieving private keys. Researchers have already successfully used quantum computing to factor small RSA key sizes. You can see this risk reflected in current guidance from cybersecurity bodies, like the Australian Information Security Manual (ISM):

So - by 2030, RSA, ECDSA, ECHDH and other cryptographic ciphers built on classical encryption will no longer be approved by the Australian Signals Directorate (ASD) for use, and we should really be looking at PQC now, while we have a lot of time to prepare. You can see an awesome article here from Clemens Lang if you want to read more on Red Hat Enteprise Linux and post-quantum cryptography.

Red Hat at Fenway park

Definitely one of the Red Hat Summit highlights this year was heading to Fenway Park to see the Boston Red Sox take on the NY Mets - with all of the Red Hat customers and partners!

Never have I seen so many red fedoras at one sporting event. It was amazing. I managed to snap a quick video here:

Sadly, the Red Sox were trounced. We had a blast though - Go Sox!

Seeing 'The Departed' filming locations

One of my all-time favourite films is "The Departed". The movie is set in Boston, and while a lot of it was filmed in Brooklyn and New York, I wanted to see if I could find any filming locations in Boston.

One of these was actually really easy. We were staying in the Seaport district of Boston, and if you head up the road less than a kilometer to 12 Franklin Street you can see an iconic scene from The Departed. I won't spoil the plot for you if you haven't seen it - but here's the scene, and the spot it was filmed.

If you look the other direction, you can see some of the green windows in this scene:

I think this scene with a fire escape was filmed on a nearby building, just around the corner:

While you're there, make sure you check out the nearby "Flour" cafe. I often find it hard to get a decent coffee in the USA, but this was amazing. Definitely worth it!

If you head further along the road to the Boston wharf district you can find another filming location. It's right at the end of the Long Wharf, past the Marriott. This section was under repairs when we got there, and if you want to check it out I would head there soon - it might not survive much longer! Here's the scene, and me doing my best to find it amongst the construction fencing (and not get "trespassed").

Other things to do in Boston

There's a few other things I would highly rate doing while you're in Boston:

Firstly, the Freedom Trail tour is iconic. It's a red-brick path that navigates you to historic sites throughout Boston, that were significant in shaping America. These included Boston Commons, the Granary cemetary, and the building where the Boston tea party was planned. I booked a tour through Expedia, and it was incredible value. Our costumed guide was friendly, funny, knowledgable, and generally awesome.

I was also recommended Ristorante Limoncello by another Red Hatter, and it was incredible. The food and service was amazing, and I really felt looked after. I had the Rosette Al Montasio Ed Olio Di Tartufo, and it was outstanding. I can't rate it highly enough.

Trailblazer award

So what was I doing in Boston for Red Hat Summit anyway? This year I was incredibly honoured to be the recipient of the Paul Cormier Trailblazer award. For those not in the know, Paul Cormier was the Red Hat President and CEO, and set an example for leadership, courage, and strength, that is still felt today. Each year Red Hatters are nominated by their peers to receive the award.

The first event on the Trailblazer agenda was a tour of Fenway Park. It was facinating hearing the history of the ballpark and the Boston Red Sox, including the background to the "green monster" and why the "monster seats" are so expensive (~ USD $700-1000 per game!).

I also learned about the "Curse of the Bambino", and saw Babe Ruth's bat.

Part of the award was a dinner with Trailblazer award winners and Red Hat leaders, including Paul Cormier himself. I had an awesome time speaking with the other Trailblazers, and hearing their stories of innovation, and creating solutions to really interesting problems. Richard Chun also graciously snapped a pic of me with the Red Hat President and CEO, Matt Hicks.

I'm incredibly thankful to the Red Hatters who nominated me; it was an awesome experience at Red Hat Summit and the Trailblazer events in Boston. A huge thank you also to Kara Strang and Kelia Pless who made the week a once-in-a-lifetime experience.

A very special thank you to my partner Beth, for all your support, and coming along and celebrating with me. You're the best :)