This presentation has been presented at the "Vienna DevOps & Security Meetup" in 2021.
It discusses the state of monitoring, what Opentelemetry is and why should you care about it.
Concepts and basics are discussed and presented in a full example extracting traces, metrics and logs.
Demo: https://github.com/secustor/opentelemetry-meetup
5. What’s out there? / The state of monitoring
- Microservices lead to a more complex debugging process
- OSS
- Observability pillars are separated
- Proprietary products
- Heavy vendor lock in, especially if SDKs are used
- Public clouds
- Mostly support only their own products
- Limited imports available
- Product changes require system redesigns
- Metadata is not standardized
6. What is observability?
- We need to answer questions about our systems.
What characteristics did the queries that timed out at 500ms share in common? Service
versions? Browser plugins?
- Instrumentation produces data.
- Querying data answers our questions.
10. But why OpenTelemetry and not Y?
- Semantic conventions for metadata
- Open standard
- Supported by :
- + 300 companies
- + 2,000 contributors
- + 100,000 contributions in the last year
- CNCF ( incubating )
- FAANG ( MAANA? 🤔) + M
19. Status
- Like Kubernetes in 2015/16
- Basics are there, advanced stuff needs some work
- Most SDKs GA
- Auto instrumentation available for Java and JS
- Opentelemetry Operator in development
- Java agent injection possible