I'm setting up monitoring for several popular SaaS applications, and I'm facing a financial dilemma with major observability services like Datadog and New Relic. They charge per 'Synthetic Run', meaning if I want to monitor a critical flow, like from 'Login to Checkout', every minute across three regions, the costs skyrocket. It seems counterproductive to have to monitor less frequently to manage costs, as this undermines the idea of real-time monitoring.
I'm curious about the experience of others in the SRE and DevOps communities: do you experience 'bill shock' from synthetic monitoring? Do you absorb those costs for critical processes, or have you created in-house solutions to avoid those vendor fees? I'm weighing whether to accept the high costs or invest in a more cost-effective solution using AWS.
1 Answer
Synthetic checks can really add up! From what I've seen, it depends a lot on your organization's needs and how critical downtime is. If you're a big player like Amazon, a 15-minute outage could be catastrophic financially. For a lot of us, though, using something like Lambda to monitor uptime can be a much cheaper alternative than staying with the big names.
That's a huge difference in pricing! When I broke down the costs for running a complex flow every minute across three regions, it looked like it could hit about $1,500 a month just for monitoring. I find it wild that keeping an eye on things could cost more than the production servers. Has anyone here developed a reliable in-house option with tools like K6 or Playwright?