21 Dec 2023 Code, Cloud, and Casinos: Surviving AWS re:Invent 2023
After a massive week of announcements, AWS re:Invent is all done and dusted for 2023, so it’s time for a recap on how it was from myself and the two other Shiners that attended, Associate Principal Engineer Doug Hagan and Senior Software Engineer Tyrus Caldeira. In this post we’ll share our experience planning for the event, our thoughts on the sessions we attended, and tips for future re:Invent attendees.
Read on to hear from each of us in turn, or skip to a specific one:
Michael
“Oh my goodness…”. This was my initial reaction when I opened up the agenda planner for re:Invent. The sheer amount of sessions to choose from was just mind boggling and it’s simply impossible to catch every session you’re interested in, especially when you take into account travel between venues.
My day-to-day role primarily focuses on designing serverless solutions on AWS, so my goal coming into re:Invent was first and foremost to seek inspiration from hearing what peers in the industry were doing, and to focus on architecture topics such as best practices, cost optimisation and resiliency in serverless applications…along with a sprinkle of GenAI.
It became apparent throughout the conference that AWS has been laser focused over the last year on bringing a GenAI product to market to compete with the likes of OpenAI. As such, the majority of sessions ended up touching on this topic and unfortunately, there weren’t many major announcements other than Bedrock improvements and Amazon Q.
On the plane over to Las Vegas, I was continuing my read of “21 Lessons for the 21st Century” by one of my favourite authors Yuval Noah Harari. It happened to touch on the technological challenges facing humankind and the shift in authority from human to machine, threatening to render a large part of the population irrelevant.
I couldn’t help but think back to this chapter whenever I was in a session covering GenAI. It is inevitably going to touch on every aspect of what we do as technologists, from designing solutions, to developing and operating them. It makes expert knowledge available in context and there is no doubt it helps engineers deliver their work more efficiently (Amazon boasts a 28% increase in productivity using CodeWhisperer).
Skilled engineers are still required for nuanced decision making but I cannot help to worry about the potentially less experienced ones that aren’t able to provide that level of reasoning required and, at risk of sounding alarmist, becoming irrelevant. How does a junior engineer gain experience in an environment where all the application of knowledge is performed by an algorithm?
Doomsday talk aside, it’s an exciting technology and AWS is drastically reducing the barrier to adoption. It’s up to organisations to assess the relevance of Generative AI within respective functions and how well businesses are able to harness this capability will become key to securing market share. However, it’s imperative to be mindful of the limitations of this technology and let’s all keep in mind the following: “Without good data, there is no good AI”.
The conference was extremely well run. The lunch service was speedy, the travel between venues straightforward and that’s no small feat given the huge crowd that re:Invent draws. It would be fascinating to peek behind the curtain to witness the logistics involved in putting together such an event. Similarly, the sessions from AWS customers at re:Invent allowed us to peek behind the curtain to see how they run their workloads on AWS and tackle availability and resiliency challenges. As a solution designer, I enjoyed seeing how financial organisations implement multi-region failover, how Amazon Prime’s team places availability at the heart of its culture, and recommendations from Amazon on fostering innovation, etc.
I concluded my week at the epicentre of the AWS cloud community feeling both drained yet energised by the knowledge I had acquired and the innovative applications of AWS that I witnessed. Industries such as aerospace, automotive, live sports broadcasting…it was all fascinating stuff. I now have about 3 years’ worth of content to catch up on, and look forward to implementing some of these ideas and innovations to unlock greater value for our clients.
Doug
AWS and Las Vegas are well suited for each other: they’re both flashy, have high production values, and also unexpected hidden depths. I went in with no expectations or goals from my day-to-day role, other than keeping my eyes and ears open. Sometimes the conference talks that resonate the most with me and inspire thought afterwards are the ones that I stumble upon without planning. So instead of having a particular focus I aimed to experience a wide variety from what was on offer, both in content and format.
Some of the standouts for me were:
- Re-skilling at the speed of cloud: Turning employees into entrepreneurs by Phil Le-Brun
This was about creating organisations that have the skills and shared vision to achieve great things. The gist was that everyone should spend 30% of their time learning, but that 70-90% of learning is done on the job. Training is important, but inadequate alone and unused learnt skills atrophy within 3 months. Furthermore, tech skills don’t stay relevant for too long, but “soft” skills like leadership are immutable. - Developing next-generation quantum networks
This was a topic that I’d had no previous exposure to, but came away from with a new appreciation for. Quantum computing sounds great in-principle. Cramming more-than-binary information into qubits seems like a good way to bust conventional speed and power limitations for computing. The way that a synthetically grown diamond with a very particular shape and intentional atomic defects can catch a photon qubit and store it was mind blowing for me.
However, for real-world applications we need memory and networking as well. Networking requires repeaters, as observation of a qubit collapses it, so the entangling of two entangled pairs can inversely associate the non-entangled pair and extend range from the 10km hard limit. I feel like I’d need another lifetime to properly understand any of this, but it was really interesting and exciting to see how fast this field is progressing. - Vendor Expo
This was just huge, probably the entire Melbourne Exhibition Centre would fit in the expo space. I spent about 6 hours over a couple of days wandering through the vendor stands, playing their games and talking to them, admittedly sometimes just for an awesome t-shirt or hat (shoutout to the red hat from Red Hat) but often because they were of genuine interest.
Everyone wants to talk about how AI has revolutionised their space: coding, infrastructure, integrations. I felt there was a lot AI can offer, particularly in the cybersecurity intrusion / fraud detection and mitigation space. The application of AI to log analysis, packet inspection and event streaming could lead to systems that can detect an attack more reliably, then react much faster to mitigate the risk. (Of course I’m sure there’s an equal amount of AI on the attacking side dedicated to improving their sophistication.) - F1 watch party: Behind the Abu Dhabi Grand Prix
This was a gems that not too many people attended (maybe because it was at 6pm) but it was a fascinating insight into the world of big data through the lens of Formula 1 race engineering. One of the things that’s really changed the way that teams experiment and innovate is the evolution of ever-more-complex digital simulations, and these are a really useful tool in many other industries as well. In the final part of the session they presented selected qualifying and race laps from a few weeks earlier, accompanied by the telemetry graphs that the engineers use to analyse and improve on the fly. They showed us how the torrent of incoming data guides the race teams to make impactful tactical and strategic decisions. - Inclusive design thinking in tech: Fostering innovation for all
This talk was a discussion of the benefits of designing more inclusively, with great examples and a set of steps to help move an organisation in the right direction. Some of the calls to action included: creating inclusive design principles, engaging early and often with user communities to avoid unconscious biases, integrating (continually evolving) accessibility standards, educating teams and diversifying the talent pool. - Renewable energy assets in the smart home: Managing demand flexibility
This was the only hands-on workshop I attended, and was about creating digital twins, which in this case were used to simulate energy scenarios between a house, an electric vehicle and an air conditioner in order to optimise power usage over the course of a day. I hadn’t used any of the services before but enjoyed playing along and putting them together, and I learnt a lot about what digital twins are and how they’re useful to tweak designs. - AWS GameDay: Formula 1
I’ve done a few AWS game days over the years. This one was F1GP themed and had us playing the role of data techs supporting the race engineers for a team owned by good old Unicorn Rentals. It was fun, and our team was varied enough in skills to work on some tasks in parallel, but unfortunately we were affected by some issues (including one that saw us temporarily climb to the top of the leaderboard through a scoring bug). It was a good GameDay to run though, covering quite a few different AWS services. - New AWS generative AI features and tools for developers and
Realizing the developer productivity benefits of Amazon CodeWhisperer
Software development has always been a tradeoff between holding onto the favourite tools that we have many years of muscle memory with (like the handle of a favourite saucepan or screwdriver), and adopting the latest, greatest thing. Today, the latest thing is, of course, GenAI. It feels like we’re at yet another AI tipping point when it comes to writing code, and that someone who comes to terms with these new tools and integrates them into their editor will subsequently be some large percentage (I heard various numbers like 30 and 60 percent mentioned) more productive and faster than someone who doesn’t. I think this gain will be somewhat dependent on what kind of code is being written, and generated code must be thoroughly reviewed for suitability and stupidity, but overall most people should see a fair amount of time saved. - Sustainability Showcase
This highlighted projects using AWS services to improve (largely environmental) sustainability around the planet. One stand that stood out to me was a prototype of the use of Rekognition to analyse images taken by webcams on firewatch towers in parks and forests. Having a system that can distinguish between fog, cloud, steam and smoke is impressive, and the speed of response that automation of analysis enables will make a huge difference to the spread and impact of bushfires. There are many possible extensions to this idea that could make smoke detection even more accurate: video analysis (movement over time), heat cameras for detection at night, correlation with other data like wind direction and speed, and it’s really exciting to see the way that the AWS services fit together to do all the heavy lifting.
The way that the casinos, hotels and convention centres swallowed over 60,000 conference attendees without a trace was smooth and well practised. Some of the logistical triumphs were feeding breakfast and lunch to many thousands in each of 5 or 6 different locations simultaneously, the (generally) smooth and punctual flow of buses shuttling attendees between locations, and the small army wearing AWS “Ask Me” shirts who assisted with navigating through the various mazes of corridors and rooms. The keynotes also had live music and amazing staging (overhead and incidental ambient screens everywhere) so that there wasn’t really a bad seat in the house.
That said, when the shuttle buses didn’t arrive quickly enough or fatigue meant the will to race to the other end of the strip had drained away, there were invariably great alternative options close by. Often sessions would be simulcast in the silent theatre overflow rooms, or there were walk-in seats available for other interesting sessions. There were also many great casual options like the Sustainability Forum and the Vendor Expo.
What I took away from this whole experience was a sense of the scale and pace at which the things we all build are evolving. ChatGPT was released just over 12 months ago and now almost every talk, regardless of the topic or audience, made some reference to the impact that Generative AI is having on our industry and culture. AWS is making it easier to build on GenAI-as-a-service with offerings like Bedrock, which greatly reduces the barriers to entry. People driving company direction want to be proactive in adoption of these new tools that sound like they could be a miracle solution, and people building things want to reduce the amount of boiler-plate and repetitive code they have to write, not to mention the reduction in Googling the next bit of syntax or new language feature. There’s an aura of inevitability around it, a sense that we’re already on the AI rollercoaster so we’d better just buckle up and get used to it.
That reminds me of the first talk I went to, which was all about creating a culture of learning in our organisations. So many of us write “life-long learner” or something similar on our resumes, but it’s at work that we do most of our learning, and we need to build that practice into our teams and our day to day actions, because every year there’s something new. I wonder whether next year it will be quantum.
Finally, here are some tips for those who attend in future:
- Everything is harder to get to than it looks on a map, casinos seem to be designed to keep exits hidden. A 30 minute gap between sessions is rarely enough if venues aren’t actually adjacent, so plan sessions at the same location for half a day at a time. Walking from casino to casino rather than exiting to the strip is often easier.
- Stay at a central hotel (not Mandalay Bay or Wynn) so that many sessions (Venetian, Caesar’s Forum) can be walked to.
- If you are Australian like me, get used to looking left for traffic, and staying to the right on sidewalks and escalators.
- Register for livestream access as well for the keynotes and innovation talks, as it’s really convenient to be able to watch things that were missed either on the go or from the hotel.
- Spend some time walking through the casinos, there’s a lot to see (especially the Grand Canal in the Venetian, and the fountains outside the Bellagio).
- If you’re all Vegas-ed out, a great place to relax is the Downtown Container Park. Good food, good entertainment from an open mic stage, and no casinos or spruikers.
Tyrus
I had a great time at re:Invent, but like Doug, there are a few tips I’d share with future re:Invent attendees:
Plan Your Calendar Before!
- Before going to re:Invent, they advise you when sessions open up for reserving seats at different sessions. This usually occurs about 2-4 weeks prior to the event. Prior to the conference, I spent about 5 hours over over 3-4 weeks looking over the entire list of session and planning what to go to. There are hundreds (or thousands?) of events and you’ll only get to maybe about 3 a day, so of the 4-5 days of the conference, you may only get to about 15 sessions or less. Don’t be stressed by this as it depends if you want to do the workshops (usually 2 hours) or other talking sessions (usually 1 hour) or GameDay/Jam Sessions (usually 3-5 hours) or keynotes (usually 2 hours)
- If you want to do the technical sessions and avoid the sales pitches, go to the 300 and 400 level sessions, not the 100 or 200 levels.
- Sessions are reserved 10 minutes prior to start time, after that its first-in-best-dressed via the walkup line
- In this year’s conference, both breakfast and lunch were provided everyday Monday-Friday and it was quite good so worth going to. But for dinner you’ll have to find a restaurant or takeaway place to go to.
- The re:Invent app is good but it isn’t perfect. Half my sessions were listed as walk-in the week before, but on Monday had changed to reserved sessions
Location
- The conference was spread around a few main venues – The Venetian, Wynn, MGM Grand, Mandalay Bay, Caesar’s Forum (not Caesar’s Palace) and Encore. The majority of sessions were held in The Venetian, Wynn, MGM Grand and Caesar’s Forum. So we stayed at the LINQ Hotel + Experience which, whilst not super-fancy, is within walking distance of the main venues. You will need to take a shuttle to MGM Grand or Mandalay though. The best part about the LINQ is the Promenade which has a strip of shops and food, which was great for going to dinner. The High Roller is also a great thing to ride as well as going to visit The Sphere!
- The venues are huge and it’s best to wander around prior to attending re:Invent to understand where the rooms will be and the best way to get there. Also, look at the official app, as it has venue maps!
- Getting between venues takes time. Sometimes 30 minutes isn’t enough, especially if you need to catch a shuttle bus or rideshare taxi. That said, don’t pay the extra for a rideshare taxi unless you’re really desperate.
Las Vegas itself
- The Las Vegas F1 race ran the week before AWS Re:Invent 2023. This caused a lot of traffic disruptions for both locals and visitors and it also made the shuttle buses take longer. Sometimes from Caesar’s Forum to Mandalay bay took about 30 to 45 minutes. So if the F1 race is the week before again for AWS Re:Invent24, then be prepared for traffic issues!
- Las Vegas is great to be at and the best part is walking through the hotels, but if you’re not a gambling person, don’t be shocked that there’s a casino in every hotel. Most hotels you walk into, you’ll end up in the casino which can be fun the first or second time, but the third and fourth etc gets a bit boring. If you want to go “shopping” at stores that don’t cost a fortune then you’ll want to move away from the strip to other shopping malls. Sadly we didn’t get to do this but we saw some on the way in. One nice and calm place away from casinos was Container Park. It’s a park made out of shipping containers with a bunch of small stores and a few places to eat and a live stage. The live stage is made up of local artists or anyone that wants to rack up the courage to sing or do a poem! It was quite fun and relaxing. It’s also a few blocks from a place called the Fremont Strip, which we thought would be full of shops but sadly it’s just more bars and casinos, but still a nice place to visit.
Conclusion
All in all, re:Invent was well worth the 14 hour flight (and associated delays) for us. We highly recommend re:Invent to anyone working with AWS that wants to immerse themselves in an atmosphere of excitement around Cloud and come back home full of new ideas, solutions to try out and connections. It’s worth noting that a huge amount of sessions are available on Youtube for on-demand viewing. We’ll certainly be going back to watch the multitude of sessions we wanted to attend but couldn’t realistically squeeze in. But for now, we’ll end with the closing quote from Amazon CTO Werner Vogel’s keynote: “With all of that.. NOW GO BUILD!”
No Comments