AWS re:Invent 2024 Keynote re:Cap

Table of Contents

• Overview
• Keynotes
  • Matt Garman
  • Dr. Werner Vogels
  • Dr. Ruba Borno
  • Dr. Swami Sivasubramanian
  • Peter DeSantis
• Conclusion

Overview

Evolvability: A Theme Discussed by Dr. Werner Vogels

The AWS re:Invent conference always has a number of keynotes, each with a different focus. This year they had five different keynotes, but I personally was only able to attend the keynotes for Matt Garman and Dr. Werner Vogels. In this post, I will provide an overview of these two keynotes and also briefly mention the other keynotes for completeness.

Keynotes

Matt Garman

re:Invent 2024: Matt Garman's Keynote

Matt’s keynote (recording) had a large amount of information presented, but I personally took note of a number of announcements of new product launches:

1. S3 Table Buckets: This new S3 feature allows users to create and manage Apache Iceberg tables directly in Amazon S3, enabling efficient analytics workloads.
2. Improved EC2 Instance Time Synchronization: AWS has enhanced the time synchronization between EC2 instances, ensuring more accurate and reliable timekeeping. This feature is being leveraged for the new DynamoDB multi-region strong consistency offering.
3. Amazon Aurora DSQL: This new Aurora offering is the fastest serverless distributed SQL database which can be leveraged by always-available applications.
4. Amazon Bedrock: Several advancements have been made for Amazon Bedrock, including multi-agent collaboration capability and enhanced data processing and retrieval capabilities.
5. Amazon Nova: Amazon Nova foundational models were introduced:
   • There are different model levels, including Micro, Lite, Pro, and Premier.
   • Additionally, Amazon Nova Reel can create six-second videos, with plans to extend this to two minutes.
   • They are also working on a speech-to-speech model for Q1 2025 and an any-to-any model for mid-2025.
   • The speech-to-speech model will interpret verbal and nonverbal cues, which will allow it to deliver natural, human-like voices.
   • The any-to-any model will power applications from translators, to content editors, to AI assistants.
6. DynamoDB Global Table Multi-Region Strong Consistency: A preview was announced for multi-region strong consistency for DynamoDB global tables. Multi-Region strong consistency ensures your applications can always read the latest version of data from any Region in a global table. This feature will be useful for things which demand strong consistency, such as user profile management, inventory tracking, and financial transaction processing.
7. GitLab Duo with Amazon Q Developer: An integration between GitLab and Amazon Q Developer was announced named GitLab Duo. This new integration leverages AI agents to assist with complex, multi-step tasks such as new feature development and codebase upgrades. With Amazon closing down their CodeCommit git repository service earlier this year, I suspect they have a plan to further integrate with GitLab as their preferred Git repository provider.

Amazon Q Features

8. Amazon Q Developer Transform Capabilities: A new feature of Amazon Q Developer was announced named Transform. This feature leverages the transformation capabilities of Amazon Q Developer, which allows modernization teams to deliver large and complex projects, accelerating .NET porting, mainframe modernization, and VMware migration, all while enhancing application security, resilience, performance, and scalability.

Amazon Q Developer

Overall there were a lot of announcements in this keynote and I am having trouble picking ones that are more important than others. There are some themes that I picked up from these announcements though:

1. Amazon is investing heavily in making it easier to create your own Generative AI solutions. The enhancements to Bedrock and the release of the Nova foundational models allude to this fact.
2. Strong consistency of data (globally) coupled with faster writes and retrieval are not only nice features for their customers, but I suspect are useful to Amazon for their global AI products.
3. There is a big push to add AI-powered tooling throughout Amazon’s service offerings. Despite this theme being a bit exhausting during the course of re:Invent, I do think AWS did a good job of presenting valid use-cases where AI did help accelerate productivity.
4. It does appear that AWS and GitLab have a growing partnership and, as I mentioned above, I am guessing that is part of the reason why AWS stopped investing in CodeCommit. I would not be surprised if GitLab is eventually purchased by AWS.

Dr. Werner Vogels

Dr. Werner Vogals always has an entertaining way of doing his keynotes (recording). There was an admittedly odd yet entertaining skit at the beginning about a pizza delivery guy (Alex) helping them solve complexity issues with S3. This skit was followed by Dr. Vogels highlighting his 20 years at Amazon and 13 years of the re:Invent conference.

He mentioned that around the time of the first re:Invent, he gave some advice on 21st century architectures, which needed to have four key features:

1. Controllable
2. Resilient
3. Data-driven
4. Adaptive

21st Century Architectures

Dr. Vogels then went on to discuss complexity, he used an interesting analogy of a unicycle, bicycle, and tricycle to help explain the balance of simplicity and performance.

Unicycle vs. Bicycle vs. Tricycle

During the course of his keynote, Dr. Vogels outlined six main lessons for managing complexity (or as he called it, Simplexity):

1. Make evolvability a requirement.
   • This point emphasizes the importance of designing software systems that can easily accommodate future changes.
   • A software system's ability to evolve is crucial for its long-term success.
   • Evolvability is different from maintainability. Maintainability ensures a system works in the short-term while evolvability focuses on how to manage complexity over time.

Build Evolvable Systems

2. Break complexity into pieces.
   • This involves breaking up a megaservice into smaller, loosely coupled services.
   • The idea is that if an engineer cannot have a mental model of the service, then it is too big.
   • This approach helps in managing and understanding the system better.

Break Complexity into Pieces

3. Align your organization with your architecture.
   • The structure of your teams should reflect the architecture of your systems.
   • This alignment ensures that each team can own and manage their components effectively.
   • A culture that encourages continuous questioning and ownership leads to better quality and innovation.
4. Organize into cells.
   • Reduce the scope of impact by defining the system into independent cells.
   • Decompose things into smaller building blocks that each work independently.
   • Cells create order and can be organized using things such as Shuffle Sharding.
   • A control plane is needed to manage the cells.
   • A cell should be big enough to handle the biggest workload that you can imagine.
   • A cell should be small enough that you can actually test it with a full-scale workload.
   • But, in reality, a cell should be somewhere in the middle of the big enough and small enough statements.
   • This approach helps reliably scale a system to a full workload while reducing potential disruptions.
5. Design predictable systems.
   • Reducing uncertainty is key to managing complexity.
   • Designing systems that are highly predictable in their behavior helps avoid spikes and bottlenecks.
   • Patterns like the constant work pattern ensure that processing times are consistent, making the system more robust and easier to manage.
6. Automate complexity.
   • The right question is, "What don't we automate?"
   • Once you determine that, automate everything else.
   • Manual input should only be required for high-judgement tasks.
   • At AWS, everyone has a security job, and automated threat detection is a key part of their strategy.
   • Use things like agentic workflows to help resolve tickets.

During his presentation, Dr. Vogels also noted Tesler’s Law and Lehman's Laws, both seem like good laws to keep in mind:

1. Tesler’s Law: Complexity can neither be created nor destroyed, only moved somewhere else.

Tesler's Law

2. Lehman’s Laws:
   • Continuing Change: Systems must be continually adapted else they become progressively less satisfactory.
   • Continuing Growth: The functional content of a system must be continually increased to maintain user satisfaction with the system over its lifetime.
   • Increasing Complexity: As a system evolves, its complexity increases unless work is done to maintain or reduce it.
   • Declining Quality: A systems will be perceived to be declining in quality unless it is rigorously maintained and adapted to its changing operational environment.

Lehman's Laws

Overall this was a very interesting keynote and Dr. Vogels six lessons are good food-for-thought. I really like the concept of starting with the question, “What don’t we automate?”. There are many great bits of wisdom scattered throughout this keynote and I really enjoyed it. For this keynote, I leave you with this last screenshot which has a quote about organizational complexity:

Organizational Complexity

Dr. Ruba Borno

I did not attend this keynote (recording), but Dr. Ruba hosted a keynote that focused on the value that AWS and its strategic partners bring to their customers.

Dr. Swami Sivasubramanian

I wasn’t able to attend this keynote (recording) due to a schedule conflict, but the focus was on how you can use a strong data foundation to create innovative and differentiated solutions.

Peter DeSantis

I arrived too late to attend this keynote (recording) due to another event going long, but Peter was joined by Dave Brown and they discussed a number of innovations across compute, security, storage systems, and AI infrastructure.

Conclusion

If you want to see all of these keynotes for yourself, there is a complete playlist available. While these keynotes can get a bit long, there is a lot of useful information packed into each of these keynotes.

AWS announced a lot of interesting and innovative services and enhancements at re:Invent 2024; a number of them were mentioned in these keynotes. I encourage you to take time to watch some of them and get an idea of the new AWS toys you can play with in 2025.

Finally, if you enjoyed this post, you may also enjoy my re:Invent planning/survival guide post.