New CDK Bootstrap and the EKS Cluster
In the AWS CDK Version v1.25.0, the CDK team added a new bootstrap template that includes new resources like IAM Role and S3 Buckets. From the AWS CDK Documentation: > The AWS CDK supports two…
AWS has announced a new region in Melbourne, to be opened to the public in the second half of 2022. This makes for 30 AWS regions, six of which are currently in development.
AWS released its first region in Northern Virginia in 2006. In the following years they gradually added additional regions, until they had 11 of them, including one in China and one GovCloud region, by the end of 2015. At that point they picked up the pace, with five regions added in 2016, and two additional ones every year since.
By the end of 2020 AWS has 24 regions, spread across all continents (excluding Antartica). They have also announced six additional regions: Osaka, Jakarta, Hyderabad, Melbourne, Zurich and Spain, bringing the total number of regions to 30 by the end of 2022.
In our recent article AWS re:Invent 2020 Day 2: AWS is coming to a data center (or pizza parlor) near you! we explored how AWS is offering new compute offerings in your local data center, your mobile carrier’s data center, or in a local zone. To quote from that article:
This also allows AWS to scale faster than by just adding new regions (which they are also doing, in India, Indonesia, Japan, Spain, and Switzerland).
These edge compute locations are generally still controlled from a nearby region. For example, Outposts function as a virtual availability zone in your AWS account and Local Zones are defined by AWS as “an extension of an AWS Region where you can run your latency-sensitive applications”.
But full-fledged regions also offer services which are not available on edge locations, such as Sagemaker, DynamoDB, CodeBuild and many others. To increase capacity and reduce latency for these services, AWS still needs to build additional regions.
This summer’s Gartner Magic Quadrant for Cloud Infrastructure and Platform Services painfully highlighted Amazon’s lead public cloud architecture. They are the only big public cloud provider with multiple availability zones in every region. Most regions have three AZs, some have even more, and very few have only two.
As a comparison, Azure has 57 regions, but only 12 of them offer availability zones. This also explains how many of their services are not multi-AZ enabled by default or don’t offer multi-AZ coverage at all.
This goes back to both a philosophical and a first-mover basis: Amazon has designed their regions from scratch with extreme redundancy and availability in mind. Because they were the first big public cloud player they could afford to do it right; they had no competitors to keep up with. Their main competitors didn’t have this advantage and had to play catch-up. One of the ways to do this is by stamping out ‘data centers’ instead of regions. They’re simply quicker and easier to build.
Contrary to their edge compute solutions, Amazon’s regions are their high capacity, highly redundant facilities that suit 99% of of their customer’s workloads. At the start of the corona crisis, when everybody was suddenly working from home and video conferencing, Amazon’s investment in these regions paid off. While some of their competitors could offer no new instances to large segments of their customer base, AWS chugged on without a hitch.
Both this re:Invent and at re:Invent 2019, Andy Jassy showed this “still early days for cloud” slide:
It shows that according to Amazon’s research, 96% of IT budgets are still spent in traditional IT. If only a small part of that moves towards AWS, they’re going to need all the capacity they can build if they want to keep performing as well as they have.
With six newly announced regions AWS is keeping up its recent pace of opening up at least two new regions per year. As the chart above shows, the amount of regions has followed a pretty linear path, with a small acceleration in 2016. It will be interesting to see if they continue at the pace they’ve followed for the last 5 years, or if the four new regions planned for 2022 are the first signs of a new acceleration.
This article is part of a series published around re:Invent 2020. If you would like to read more about re:Invent 2020, check out my other posts:
I share posts like these and smaller news articles on Twitter, follow me there for regular updates! If you have questions or remarks, or would just like to get in touch, you can also find me on LinkedIn.