It’s re:Invent season. We are discussing AWS re:Invent 2023, the annual AWS conference run in Las Vegas. And it just so happens that Dave has completed a talk this year for G-P.
Observability needs to be consumable for everyone.
There have been many serverless updates and announcements on Observability and the Testing Framework Beta for serverless testing. Observability is a big one. We notice that for distributed systems, observability is king. A lot is coming in about tying observability together and being less dependent on specific tools. It’s something that AWS is enabling and facilitating.
A lot of work is going on in OpenTelemetry within the community and for teams adopting serverless architecture and serverless workloads. Trace propagation can be an issue and a challenge. People need clarification when you break Observability down into dispute tracing, logging, alerting and metrics. It needs to be consumable for everyone.
Generative AI is everywhere.
Gen AI is everywhere. PartyRock came out. You don’t need an AWS account; you can do a quick app and pump in a prompt in an LLM model. Removing barriers to entry and demystifying some of that stuff is essential. PartyRock is a powerful capability. So it’s great to experiment without standing up an AWS account and spending lots of money on LLM costs.
Let’s give Mark Craddock a shout-out for his Wardley Map in February on Prompt Engineering when he pointed out that it’s not about LLM. It’s about product engineering. PartyRock makes prompt engineering easier by writing an app around it. It will be interesting to see developer advocates experiment with drawing architecture diagrams generating and stitching code together. If there is anything formal from AWS to close the developer loop from idea to production-ready code, leverage AI capabilities to speed up the feedback loop. Code is a liability.
It’s all happening in the EDA space.
Many more things are happening around EventBridge, Step Functions and additional features. EventBridge has other metrics. EventBridge is an essential EDA capability within AWS. With decentralised event-driven architectures, there are lots of opportunities for advancement. Step Functions can drive from an error state, which is significant for resiliency and EDA.
We need better guardrails or faster feedback loops for good practices. Well-architected, Trusted Advisor, Security Hub, Resiliency Hub and Compute Optimizer are becoming more consumable and timely, like faster feedback loops for developers to guide them along their journey. It’s come a long way in the last year or two.
There’s a big trend around heuristics, Resiliency Hub, Security Hub, and non-functional hubs. We could do with a Gen AI Well-architected hub where you can monitor better to make it easier for admins to ensure everything works well. AWS is democratising expertise that you used to acquire by spending time reading white papers. It’s more approachable and available.
Developer enablement and tooling
Developer enablement and tooling are massive. It’s always been a criticism that developer experience is challenging. AWS started to invest in that a few years ago. We need Serverless Land enhancements, testing practices and guidance on testing serverless for EDA architectures. The Serverless Testing Kit came out in beta. It is one thing that developers struggle with. Even though there is guidance about local testing versus testing in the cloud and changes in the testing pyramid, we need more tooling, guidance, advice and capabilities to make testing easier.
What about enablement tools for productivity hacks? We need productivity hacks aimed at developers, not like DevOps Guru or CodeGuru, but something LLM’ish. Observability or Generative AI are grounded in engineering ways of working and principles. Teams are starting to ask for good traceability through a distributed workload, and what things do they need to look at from a non-functional perspective when dealing with AI-based capabilities? In other words, these are the things that you need to stand up at scale in a production-based scenario. We need to know how teams and orgs have solved those issues.
Acquiring the AWS expertise you need
The AWS site has a mountain of knowledge with white papers, blogs and labs. They need to pump that into a model and facilitate people to ask questions, so it answers your question and gives you a couple of references. People have pulled together capabilities themselves. But we need a formal AWS offering.
There are announcements like the Hyundai partnership with Amazon.com, selling cars and disrupting that vertical. They’re going to partner up on their infrastructure stack as well.
AWS re:Invent 2023 was an exciting event. We’re planning a post re:Invent episode to review the announcements and see what companies AWS are disrupting.
