AWS launched a couple of services which gets all the attention in the press. Like Kubernetes (EKS), Fargate, Sumerian and Sagemaker to name a few. However, there were also “smaller” features launched and announced, which might be just as important or even more important than we think. Customers and partners were waiting for these features, and prior to these introductions true heavy lifting was required.
An AWS region consists of multiple datacenters completely seperated and offers high availability. Lot’s of customers however, would prefer to keep data locally for better latency (user experience performance), or compliancy and regulations. Until the Cross Region VPC Peering was launched, customers need to build and maintain their peering connections between regions themselves. Now it’s offered as a service, which included better security and saving costs. Security because data is encrypted and anonymised. And lower costs, because data is charged as inter-region data transfer costs. Although it’s currently just available in US and Ireland, we can expect support for more regions in the upcoming months.
In 2017 AWS started to support more and more cross region features. With DynamoDB you were able to replicate data between regions, but it wasn’t seamless. With DynamoDB Global Tables however, this is automated. Be aware though, that it’s eventually consistent. You still write to one region and it could take some time before the update is replicated. This feature is mainly to get better performance and multi-region fault tolerance, without the heavy lifting setup. This feature is awesome if you want higher availability for global apps which need a central location to store (meta)data.
The time having a central database cluster, managed and maintained by ourselves, is behind us. A few years ago RDS was launched with database systems like MySQL, PostgreSQL, MS SQL Server, and Oracle. Recently AWS developed their own Database System called Aurora, a heavy optimized version of MySQL and PostgreSQL. With RDS, customers were able to scale the read replicas to optimize read performance. With Multi Master it’s also applicable to scale up the writing performance. With RDS and Aurora, still servers were something customers were concerned with. It was required to scale manually, configure the cluster setup etc. Aurora Serverless is a simple, cost-effective option for infrequent, intermittent, or unpredictable workloads, because it automatically starts up, scales capacity to match your application’s usage, and shuts down when not in use.
AWS protected your servers by providing a service called Virtual Private Cloud (VPC). New services however, were connected with the internet. Private endpoints were launched for S3 and DynamoDB to fix this problem. But these were just two services. Customers can now privately access AWS services from their Amazon Virtual Private Cloud (VPC), without using public IPs, and without requiring the traffic to traverse across the Internet. Currently still limited services are available, and support for additional AWS services will be added in the coming months.
Amazon API Gateway is a fully managed service that makes it easy for developers to create, publish, maintain, monitor, and secure APIs at any scale. API Gateway is commonly used for Lambda based Serverless applications. To use API Gateway for private VPC based applications, like a Docker platform, it was quite complex to setup. With the availability of Private VPC Integration, it becomes much easier.
Systems which depend on accurate date and time, so all seriously taken systems, require time syncrhonisation with NTP. It was a bit strange this simple service still requires NTP access to the internet. EC2 instances can access this service at a universally reachable IP address.
With these seven new introductions, it becomes much easier to manage business critical applications with just a few minor steps. Improved integration with AWS services, easy connectivity and data replication between regions, and synchronized systems. I think this is awesome and one by one incredible features to implement before the big announcements are generally available. To read all re:Invent announcments, check out this page