Our takeaways from AWS re:Invent 2018
Although AWS re:Invent 2018 is now over, the announcements will have a wide-ranging effect on how we build our products over the coming years.
We use AWS for our Serverless microservices built using the awesome Serverless Framework, so it is really great for us to see the rapid iteration and improvements to the backing services that we use on a daily basis (check out our post ongoing serverless).
Here is our list of product enhancements and concepts that really got us excited.
AWS Timestream
What it is: A fully-managed time-series database
Why it matters to us: We have been looking for an inexpensive and redundant time series database for our donation system for the last month. Internally we have been using InfluxDB and Prometheus, however with our push to try and manage as few of these services as possible, this is a real game changer. We had settled on using AWS managed Elasticsearch for our real-time reporting for the donation system and had created a proof of concept, which was costly for a fault tolerant cluster, however, Timestream changes everything by introducing the same pay as you go model that we are used to in the Serverless eco-system.
Auto Scaling for DynamoDB
What it is: An on-demand approach to DynamoDB where you don’t have to pay for read/write capacity
Why it matters to us: We have previously used DynamoDB for a couple of campaign applications, it was a bit of pain to have to set up CloudWatch alarms to scale read-write capacity. This will really make us re-evaluate when and how we use Dynamo DB.
Lambda Layers
What it is: Ability to build base layers that can be used across multiple Lambda functions
Why it matters to us: We currently implement a Lambda Wrapper to share code across our Serverless projects and implement monitoring such as IOPipe. This could be a really good way for us to get code re-use outside of using NPM and wrapping all of our Lambda functions.
The Power of Voice
What it is: Voice is not limited to Amazon Echo
Why it matters to us: This wasn’t really an announcement, but more a shift in thinking for us. Amazon CTO Werner Vogels talked about how voice was used to assist rice farmers in Indonesia to know when was the best time to fertilise their crops, they had no access to smartphones and used a phone number to provide their location and were then provided with the optimum fertiliser requirements for their location, leading to a massive reduction in fertiliser use.
This thought shift has wide implications for some of the charities that we support, who are operating in areas where technology access is very limited, and voice technology is much more inclusive in that regard.
WebSocket support for API Gateway
What it is: The ability to use Web Sockets with API Gateway
Why it matters to us: Bi-directional interaction between client and server will allow for some really cool customer experiences for us and means that we can avoid long polling of our services, reducing the un-necessary load.
AWS Amplify
What it is: A deployment and hosting platform for serverless backed applications
Why it matters to us: We currently host all of our React Single Page Applications using S3 for storage with CloudFront in-front. AWS Amplify could provide us with a tidier approach to implement this, including preview branch environments which we currently stitch together with a combination of Netlify and CircleCI.
Working backwards product approach
What it is: How Amazon approaches building it’s products
Why it matters to us: Werner Vogels talked about an approach called “working backwards” that is widely used at Amazon (see Werner’s blog post from 2006). The basis of the approach is to work backwards from the customer, rather than starting with a product idea and trying to bolt customers onto it as an afterthought. The basis of this is creating a Press Release, writing a FAQ, defining User Interaction and writing the user manual before building the product. Once the project moves into development, the previously mentioned parts can be used as a guiding light for all involved. This approach really keeps the product development focused on achieving its core customer benefits and not building out stuff that takes a long time to build, takes extra resources to maintain, and provides very little customer benefits.
Serverless mentions at AWS
What it is: The large number of Serverless mentions at AWS re:Invent
Why it matters to us: Serverless was by far the most mentioned technology at re: Invent this year with over 3,500 mentions. This is really positive for us, as reaffirms the technology choices and commitment that we have made toward Serverless over the last two years.
Serverless Well Architected as a service
What it is: Self-service Well Architected reviews
Why it matters to us: As part of releasing a new core product, we will generally go through the AWS well-architected process with one of our solution architects to ensure we are building to the best standards. It will be a real sea change for us to be able to also continuously ensure this across all of our applications.
