Just like last year, we embarked on a transatlantic journey to attend the AWS re:Invent 2022 conference this year too. RST and Trans.eu's software architects used the trip as part of their team-building exercise in addition to participating in the event itself. We were represented by a team of five: Krzysztof Bąk, Tomasz Duziak, Karol Kaczmarek, Andrzej Lewandowski i Łukasz Wróbel. Most of the team took a few days off to visit New York, Washington DC and San Francisco before the conference week. The schedule was pretty hectic at times, but it was definitely worth it!
It was the first conference in Las Vegas since COVID-19. Last year, fewer people attended, vaccination certificates were required everywhere, and masks were required to be worn even at the conference itself. Presumably because of the lifting of restrictions, the number of participants more than doubled from approximately 20 to 50 thousand! And it showed – the lobby of the venue was packed. Joining sessions without reservation was short of impossible. Possibly due to overcrowding (or perhaps a sign of a looming economic downturn), the AWS gifts for certified members were less impressive this year!
The increase in the number of participants was very noticeable. I personally found it a bit overwhelming, and it made it hard to attend interesting lectures / workshops.
You can clearly see AWS's strong focus on promoting the different integrations of its services and ways to do it. For the first time I had the opportunity to participate in AWS Jam. Together with Andrzej, Karol and Trans.eu’s Szymon Banas, we gave it a shot and placed 6th out of 40 teams. I consider this a great result considering that some participants had already participated in it prior, during the morning session. This edition's takeaway – AWS pipelines are very complicated and unclear – avoid if you can… ;)
“The world is asynchronous” was the motto for one of the main keynotes at this year's AWS re:Invent. This is how Werner Vogels, Amazon’s CTO, referred to how we should look at the applications we develop – both in terms of communication (message broker / events) and resource handling.
Talking about AWS, he mentioned his architectural principles where one of the assumptions is to build distributed applications, resistant to failure with local responsibilities.
The scale of this year’s AWS re:Invent conference exceeded the expectations of AWS themselves. ;) There were twice as many attendees as a year ago, which resulted in crowds practically everywhere. You had to be on the spot at least 10 minutes before each event to secure a seat. This was a ruthless exercise in punctuality.
re:Invent 2022 was divided into several blocks, just like it was a year ago. There were keynotes, workshops, hands-on sessions, where topics were presented and discussed (particularly great if you already have experience and want to discuss specific solutions), as well as lectures.
When it comes to the subject of the conference, there were a lot of thematic meetings on individual AWS services – how to do something better, more efficiently, and cheaper.
Unlike last year, when the focus was put on “low code” technology, this year, the trending topic was migration to the cloud, and the conference brimmed with very valuable speeches and conversations allowing you to exchange experience, or learn how bigger players migrate.
Another frequent theme was AWS Amplify, where, in my case, each workshop ended with the implementation of an application that realized the objectives from the workshop.
Of course, AI could not be missing. Different AI tools were discussed in terms of potential benefits for humanity, e.g. in predicting afforestation and climate change.
Quantum computers were the next big thing at the conference. However, at the moment, apart from the fact that this topic makes your imagination go wild, we have yet to see the emergence of computers that will fully live up to the expectations.
In my recap of the last year's edition, I said “AWS can do events”. This year, the scale exceeded everyone’s wildest expectations, including mine. While last year it seemed to me “wow, it’s a huge conference”, this time there were three times more participants. I’m simply out of words to describe it.
As for the workshop part, I tried to keep a balance between sessions focused on my “private interests” (quantum computers), workshops on tools (Kubernetes, Karpenter), and gamification “jams” that test your knowledge (or lack thereof) of various AWS services. And I got what I wanted. Goldman Sachs’s rep., during their presentation on the use of quantum computers to analyze financial markets said that they are “researching the subject”, but the devices they have at the moment, which they call “quantum computers”… are not bona fide computers just yet. ;)
Workshops, as usual, were top-notch. Unlike during the last year's edition, this time we had access to properly working Wi-Fi most of the time.
Then, there were “jams”, something I'd been waiting for the most. And that was by far the most enjoyable part of re:Invent 2022 for me. This time some of the tasks focused on services such as AWS CodeCommit and AWS CodePipelines (which I highly recommend). Our 4-person team could definitely do better (we couldn't complete only one of the 12 tasks). Still, not too shabby. ;)
This year I decided to skip all sorts of presentations that can be viewed online and, instead, spent most of my time on workshops at level 400. I wanted to attend a workshop focused on Karpenter. The tool allows running several hundred pods and observe the cluster expand with new nodes that are not only optimally matched with the requirements of the pods, but also join the cluster within seconds. I recommend anyone who has had any contact with administering a Kubernetes cluster to research this tool.
At another workshop, which was focused on building a SaaS application using Amazon EKS, I managed to organize my knowledge and see how you can combine different architecture models (silo, bridge, pool) within one application using EKS. An interesting fact I learned at the workshop is the ability to create dynamic IAM policies that can be used as one of the solutions to increase isolation at the data level in SaaS applications. The holistic approach to the subject was also a big plus. I also liked the use of Istio as a service mesh to ensure isolation between users and applications.
This year, AWS did not present as many groundbreaking services as before. Instead, the conference was more about optimizing and adding value to the existing service chain, as well as presenting the achievements in the area of sustainability.
Due to the number of participants and the limited number of seats, it was difficult for me to sign up for every session I wanted. Quantum computers were one of the things I wanted to hear about. However, I was a bit disappointed – instead of a portal to another world, I saw a tool which the financial industry can use to solve optimization problems with rapid convergence. I was also presented with the limits of the human mind: with the increasing number of qubits in quantum computers, quantum software design will have to be handed over to machines. The complexity is getting so high that it is increasingly difficult for people to create programs that are not only correct, but also optimal.
I liked the common-sense trend of promoting value maximization from existing services and the connections between them. Lambda and EventBridge were presented as tools that can reduce the overhead of deploying applications to multiple regions at once. SageMaker was demonstrated as a tool for analyzing data collected from the IoT. For sure, it’s interesting and practical, but mostly just reasonable. Nothing “crazy” or “groundbreaking” about it. Perhaps this is what the near future will hold for the industry?