AWS re:Invent kicked off virtually with registration free and open to everyone for the first time this year. Andy Jassy’s much-anticipated keynote, as well as the Partner Keynote with Doug Yeum, were the featured content for the first week. Both stressed the increased need for rapid iteration and transformation.
With the announcements of ECS and EKS Anywhere, AWS itself has seemingly begun to transform from attempting to be all things to all customers, to having a more multi-cloud approach. These two services are in the same vein as Anthos on Google Cloud Platform and Arc on Microsoft Azure. All of these services allow customers to run containers in the environments of their choosing. AWS announced S3 Strong Concurrency, which also brings them in line with GCP and Azure. Andy Jassy did make sure to differentiate Amazon from Microsoft by specifically calling them out as incumbents that customers ”are fed up with and sick of.” This was part of the announcement for Bablefish, an open-source MSSQL to PostgreSQL translator.
Approximately 30 products and features were announced in these two keynotes, but the one that will impact almost every AWS customer is EBS gp3. Gp3 allows you to “provision performance apart from capacity.” On gp2, you can get increased performance by increasing the size of the volume. By switching EBS volumes to the new 7th generation gp3 volume types, customers can provision IOPS separately and pay only for the volume size they need. These new volumes are apparently faster and cheaper than gp2 in every way. Most customers are planning to switch all supported volumes to gp3 as soon as possible, but there are still some concerns about support for boot volumes.
Another important storage announcement was io2 Block Express volumes, which can provide up to 256K IOPS and 4000 Mbps of throughput. Some customers have been waiting for a cloud storage solution that can compete with Storage Area Networks they have used in on-premises environments, but, as AWS critic Corey Quinn pointed out, two of these volumes transferring at that throughput across two Availability Zones would cost somewhere in the neighborhood of a dollar per second.
Amazon SageMaker Pipelines, Feature Store, and Data Wrangler (not to be confused with aws-data-wrangler by AWS Labs) were also announced. These tools will be welcomed by companies that need to regularly clean data, store/retrieve associated metadata, and consistently re-deploy Machine Learning models. QuickSight Q was demoed and seems to be a natural language processing marvel. It can retrieve QuickSight Business Intelligence without pre-defined data models.
AWS Glue has been enhanced with materialized “Elastic” views that can be created with traditional SQL and will replicate across multiple data stores. The Elastic Views are serverless and will monitor source data for changes and update regularly.
The big Partner Keynote highlight was the introduction of the Amazon RDS Delivery Partner program as part of the AWS Service Delivery Program. Now customers can easily find partners with the database expertise to ensure Disaster Recovery, High-availability, Cost Optimization, and Security.
There is a lot more to come in the following weeks of re:Invent, and we look forward to doing deeper dives on all of these announcements here on our blog and in our podcast, Cloud Crunch! Check back next week for a Week 2 recap.
-Joey Brown, Sr Cloud Consultant