Unveiling the Future: Day 3 Highlights from AWS re:Invent 2023

As we bid adieu to the third day of AWS re:Invent 2023, our dedicated experts—Mark Noorman, Tom Eigenraam, and Hugo van Duijn—continue to spearhead the exploration of cutting-edge breakthroughs and innovations. This daily briefing serves as your virtual portal, offering a firsthand account of AWS re:Invent 2023. Whether or not you're physically present, we ensure you get updates from the ground.

New feature: According to researchers most large companies are using or looking to use a multicloud approach

What: AWS has extensive experience in aiding customers with the implementation of multicloud strategies. However, these implementations are not always undertaken for the right reasons and may not invariably achieve success. Organizations undertaking multicloud initiatives, should be aware of the risks involved. AWS is committed to sharing knowledge and lessons learned on this subject, providing valuable insights to enhance the likelihood of succesful multicloud adoption.

Why it is interesting: Once you start implementing your workloads in multiple public clouds, it will take a lot of effort to go back, so you better make sure to do it for the right reasons. General rule is that is must be driven by business benefits. One of the wrong reasons is to prevent vendor lock-in - CSP’s differ on several points so solutions are never interchangeable between them, they will always require rework. As a result you will never reach full discount potential in both clouds so in total it will be more expensive. One basic recommendation is to choose one primary cloud for most workloads, and if specific workloads are better hosted somewhere else, let that be the exception to the rule. Bottom line is that each company needs a clear strategy on which CSP’s to use that covers all the right arguments, risks and benefits, so everyone is aware of the direction the company is heading.

New feature: VPC Lattice for securing and simplifying inter-account connectivity between applications

What: Introduced last year, VPC Lattice has recently undergone enhancements designed to expedite the establishment of connectivity between applications hosted in distinct AWS accounts. Through the predefinition of services, target groups, and authentication measures, application owners now have the ability to approve connection requests. With this approval, Lattice autonomously generates all requisite network resources, eliminating the need for repeated engagement with a cloud network engineer for each instance.

Why it is interesting: Zero trust approaches are gaining popularity among companies aiming to increase their security posture and mitigate the risk of breaches. While restricting network access to resources is a common practice for heightened security, it often comes at the cost of slowing down development speed. VPC Lattice alleviates this burden by automating the process. Users can define access policies, and AWS orchestrates all the underlying services. For instance, additional authentication steps can be effortlessly integrated to verify the request origin. This is particularly advantageous when dealing with applications spread across different accounts, simplifying the process for application owners who can swiftly approve requests, leading to the automatic configuration of all underlying services.

Workshop: Bringing workloads together with event-driven architecture

Why: In a classic non event-driven architectures components usually are tightly coupled and depend on each other. This makes failures in one component a risk for the complete environment and changes to components usually impact the complete environment.

How does event-driven architecture solve this problem?

With event-driven architecture a event bridge is being used and components are configured as consumers and suppliers of events. These consumers and suppliers have no direct connection to each other which makes is a lot easier to create additional components without impacting the existing components. With placements of SQS queues between the components and the event-bridge the system can easily be made much more fault tolerant and scalable.

Bonus tip: When components live in EKS then it is advised to look into KEDA ( https://keda.sh/ ) , this is an event based autoscaler for Kubernetes so scaling can be based on SQS queue depths.

AWS Partner Keynote

As expected, there was a significant emphasis on generative AI during the event, with a particular focus on partner and customer cases delving into this topic. A noteworthy development is the introduction of the GenAI CCOE by AWS, providing partners with a platform to initiate their AI/ML-related practices. Immersion days dedicated to related services are also now accessible. While there were no updates on the Migration Acceleration Program or other funding initiatives, the keynote prominently featured advancements for Partner Central.

In a strategic move, AWS has consolidated and redefined the AWS Competencies, Service Ready, and Service Delivery programs as AWS Specializations. This restructuring includes novel specialization paths catering to Resilience, Cyber Security, and Built-In (AWS-approved IaC for partner solutions). The keynote also brought forth numerous updates on AWS Marketplace, underscoring AWS's commitment to empowering partners to scale their businesses through this channel. Integration enhancements with Partner Central allow for seamless product listings, while data-driven insights on both ACE and Marketplace business are now at partners' fingertips. Additionally, AWS has taken a proactive step by lowering pricing for Marketplace services, further incentivizing partners to leverage this avenue for business growth.