Hi, welcome to another weekly update!
Observability without breaking the bank. In this post, I took a look at Thundra’s new sampling feature and how they can help prevent observability costs from spiralling out of control. Besides the rudimentary sampling methods such as by count and time interval, they also support more intelligent modes of sampling. Such as sampling invocations that are performing poorly or erroneous.
Canary deployment with LaunchDarkly and AWS Lambda. In this post, I introduced the basics of LaunchDarkly and how you can use it to implement canary deployments for AWS Lambda. I also discussed the limitations that you need to consider, such as the number of server connections you need. Fortunately, there are ways to work around these limits. For example, by using the LaunchDarkly-relay to cache feature toggles in DynamoDB. Here‘s a CloudFormation template to run the relay with AWS Fargate.
Since the last update, I gave the opening keynote at ServerlessDays Zurich. Where I gave a brief history of the evolution from on-premise to serverless and my views on where we are headed with serverless. The slides are available here:
I also had a great (and busy!) time at DevExperience in Iasi last week. I had two talks, a panel, a short interview and a workshop in two days… But it was worth it, and the organizers definitely went above and beyond to looked after us and cultivated a nice, sociable environment.
Craft Conf (Budapest, May 7th) – I’m running a one-day “Production-Ready Serverless” workshop. Where we will cover the basics of AWS Lambda and the serverless framework, API Gateway, Kinesis, testing, CI/CD, configuration management, security, logging and tracing. You can register here, hurry though, there are only 3 spaces left.
I can also provide this workshop as in-house training and can do a more in-depth two-day version that gives you more hands-on experience. Details are available here.