AWS re:Invent 2023 has continued the tradition of being the most happening cloud computing technology event of the year. While there were many sessions covering leadership, partnership, technology updates, case studies, and much more — this article focused on sharing key takeaways from the event for software architects. Read about 2022 re:Invent takeaways by clicking here.

#1 — The frugal architect with cost awareness and sustainability mindset

Werner’s keynote has always been the most inspirational because of his candid, simple, pragmatic & effective way of communicating as per current circumstances. The key theme of Werner Vogels’ keynote highlighted the need of the hour to be cost-aware and sustainable at every step of the system architecture, design, and implementation with principles such as:

Additionally, these quotes used by Werner are a great reminder to keep looking forward and be aware of the current market situation:

Find the dimensions you’re going to make money over, then make sure that architecture follows the money — Cost-aware Architecture (2012)

Cost is a close proxy for sustainability.
To predict the future, observe the present.

The most dangerous phrase in the English language is: ‘We’ve always done it this way…’
— Rear Admiral Grace Hopper

#2 — Generative AI is pervasive, and optionality offers choices as per context

Amazon Bedrock and Amazon Q have been the key highlights of the conference with a focus on Generative AI strategy and roadmap. Adam Selipsky (CEO, AWS) and Swami Sivasubramanian (VP, Data & ML Services) have shared AWS’s perspective towards Generative AI for enterprise:

• Amazon Q (Generative AI assistant) launched as part of the event and was highlighted by industry researchers as one of the most strategic steps by AWS considering the competitive enterprise offerings such as Google Cloud’s Duet AI and Microsoft’s Copilot stack.

• Siva focused on providing the optionality as part of Amazon Bedrock to choose from the diverse set of foundation models. It is similar to AWS’s RDS strategy to provide options, the approach has been to provide choices to their customers to choose the suitable database technology. Click here to read more from my perspective on current features by Amazon Bedrock.

We are a strong believer that no model will rule the world because there are absolutely going to be different models for different use-cases. You need a model choice and it is paramout.

— Swami Sivasubramanian (VP, Data & ML Services)

Model Evaluation feature also launched as a Preview feature:

Other Key announcements:

• Amazon Bedrock as a middle layer supporting 20+ models — launched Claude 2.1 & Lllama2 70B models.
• Amazon Titan Multimodal embeddings for search, recommendation, and personalization. Titan image generator launched with an invisible watermarking feature.
• RAG and fine-tuning and customizing base models through Bedrock-supported models.
• Launched Knowledge bases for Bedrock, which supports full-managed RAG and store embeddings in supported vector databases.
• AWS Generative AI CDK Constructs with sample implementations for common generative AI patterns. Click here to know more.
• AWS Cost and Process Optimization with Graviton-4 (Preview) — Graviton-4 is up to 40% faster for databases, 30% faster for web applications, and 45% faster for large Java applications than Graviton3.

#3 — Data engineering is the foundational building block and drives the ecosystem

The focus on data engineering is evident as it not only drives the applications’ ecosystem but also is the foundation for building AI-related capabilities. With Snowflake and Databricks leading the market share, AWS has a multi-dimensional approach with a “build, buy, or bring-your-own” data engineering platform. Key announcements as part of the conference:

• Amazon S3 (one of the most commonly used storage for data lake) launched Express One Zone — which will be a game-changer technology for data lake because of its availability (99.95%), cost, and performance efficiency. Data access speed can be improved by 10x and cost can be reduced by up to 50% in comparison to the S3 standard. Click here to read more.
• Announced Amazon Redshift Serverless with AI-driven scaling and optimizations. Click here to read more.
• Vector database capabilities available with RDS (PostgreSQL), OpenSearch, Aurora, DocumentDB, and DynamoDB — it will continue to expand with other database products. Note that you can also bring-your-own vector database with Bedrock’s knowledge base feature:

• For building a better data foundation, Zero-ETL has been the key focus and support between Aurora & Redshift, DynamoDB & OpenSearch service, and OpenSearch.
• Launched anomaly detection with AWS Glue (Preview) to improve data quality by using ML to detect statistical anomalies and patterns. Click here to read more.
• AWS Clean Rooms with enhanced features — While it helps to share data with partners, it can be leveraged for sharing data in a controlled way with enhanced privacy protection (differential privacy). Click here to read more.

References:

• Top announcements of AWS re:Invent 2023
• The Frugal Architect by Werner Vogels
• AWS Events Channel for AWS re:Invent 2023 Session

Disclaimer:
All data and information provided on this blog are for informational purposes only. This makes no representations as to the accuracy, completeness, correctness, suitability, or validity of any information on this site and will not be liable for any errors, omissions, or delays in this information or any losses, injuries, or damages arising from its display or use. All information is provided on an as-is basis. This is a personal opinion. The opinions expressed here represent my own and not those of my employer or any other organization.