Afternoon! We'll keep this quick since it's a holiday here for us and much of you as well. First, some thoughts...
Synchronous Versus Asynchronous Monitoring
This week we have links to posts from Thundra, IOpipe, and Dashbird; all competitors in the serverless monitoring space. Right now the biggest divide in the serverless monitoring space is synchronous versus asynchronous monitoring. How this plays out is in what service is responsible for shipping data to the monitoring platform. Should the data be shipped by the Lambda function or sent to CloudWatch first and ingested from there?
Both of these approaches have their own benefits and drawbacks. Both of them have different situations where they'd be preferable.
I'd argue synchronous monitoring is more important in a dev environment. You may be asking yourself, "Shouldn't the speed of alerting be more important in production?" Yes, but, you probably don't respond fast enough in production to make the decrease in alert delay actually worthwhile. Synchronous monitoring makes sense in a dev environment because of the small scale of events, the need for a short development feedback loop, and the fact you are often actively staring at the monitoring system in question waiting to see if there is an issue.
Simple request, don't tie your platform to one or the other. Make this this something we can toggle.
I'm Tom from ServerlessOps and we provide services to make you successful with your DevOps transformation and enhancement through AWS serverless adoption. Ask us about our training and advisory services.
Also let us know what you think of this week's Cold Start using the 👍/👎 links at the end of this email!
Tom at ServerlessOps