Reducing Observability Costs: How to Save $80k a Month

Hey fellow devs!

Let’s talk about something that kept many CTOs awake last year: the massive bill from observability vendors. We all love our dashboards, but at what cost? Today, I want to share a real-world example of how migrating to an open-source stack can save huge amounts of money.

The “Tax” on Every Request

Vendor pricing models can be complex. You often pay per user, per gigabyte of data ingested, and even per query you run on your own dashboards. As your system grows, these costs can spiral out of control.

Thiago shared a shocking (but common) scenario from his experience:

Thiago: We have tons of microservices running Java, and we were using New Relic and Datadog. We were burning almost $80,000 monthly in licensing and data ingestion. You pay by users, you pay by the amount of data you ingest, and you even pay by the number of queries you do on your dashboards.

The Open Source Alternative

By switching to an entirely open-source stack hosted by themselves, they were able to eliminate those massive licensing fees. Their stack now looks like this:
– Grafana Tempo: For traces.
– Grafana Loki: For logs.
– Grafana Mimir: For metrics.

Democratizing Your Data

Beyond the money, there’s another huge benefit: Democracy. When you pay per “seat” (per user), companies often limit who can access the observability tools to save money. This means developers often have to ask someone else to check the logs for them.

By using an open-source stack like Grafana, you can give access to everyone. As Thiago mentioned:

Thiago: Now with OpenTelemetry and Grafana, we can give anyone access. The access is based on our enterprise Gmail. They just use that to log in and have access to all the dashboards. It’s more democratic. You can give access to anyone without thinking about costs.

Key Takeaway

Observability shouldn’t be a luxury. By using standards like OpenTelemetry, you can build a powerful monitoring system that scale with your traffic, not with your budget.

Is the cost of monitoring holding your team back? Let’s discuss in the comments!

Happy coding!

Cheers,

Pedro Cavalero