AWS re:Invent 2023: Day 2 Announcements

Day 2 was action packed with many great announcements unsurprisingly mostly AI related!

Zero ETL Integrations: Simplifying Data Movement

Zero ETL integrations are now available between common AWS managed databases and Amazon Redshift. This is kind of a big deal as it removes the need for building out custom ETL pipelines when moving data from your online transaction processing systems to your data warehouse for analytics.  Near realtime ETL pipelines are tricky to setup and maintain so this is some very awesome work by AWS to remove the undifferentiated heavy lifting for these tasks.  Only the Aurora MySQL engine is GA with Aurora PostgreSQL, RDS MySQL and DynamoDB all in preview.

Amazon Q Launches

AWS’s answer to Microsoft’s copilot was unveiled with Amazon Q which is a conversational AI powered assistant.  It also has a number of integrations throughout the AWS ecosystem.

Amazon Q can be used for businesses, to pull together knowledge about the business from various data sources and provides a query interface to ask business questions rather than manually searching through documents, databases or other enterprise systems for the answers.

This look on the surface to be a very powerful tool.

Other integrations are for developers, QuickSight, and Connect. Check the links for more details!

Specifically around the developer side of things, Amazon Q has been trained on 17 years of AWS knowledge, the integrations include being accessible from within the AWS console, helping with determining root cause for error messages, optimising EC2 instance selection, network troubleshooting, an extension for supported developer IDE’s via the AWS Toolkit extension, and what looks to be a very powerful ability to upgrade applications with Amazon Q code transformation with Java the supported language initially.

Amazon Q is in preview for now but definitely worth a look!

Amazon S3 Express One Zone: Speeding Up Data Access

Amazon S3 got some love with Express One Zone storage class being released which allows much faster (single digit millisecond latency) access to data stored in S3 with the tradeoff of lower durability and lacking some other features.  Very conveniently this will have many use cases for AI and ML.  Only available in limited regions for now but definitely worth a look.

Graviton4: Powering Ahead with ARM Chips

Finally Graviton4 the new version of AWS’s ARM chips have been released which provides up to 30% better compute performance than the previous Graviton3 chips.

Amazon Bedrock: Fine-Tuning and Continued Pre-Training

Finally, on the Amazon Bedrock news, fine tuning and continued pre-training in Amazon Bedrock is available in preview. Bedrock now allows organisations to train supported foundation models on their own unlabelled data so they can return more specialised domain knowledge answers for your organisation. For now available only in limited regions.

Enjoyed this blog?

Share it with your network!

Move faster with confidence