Mar 27
Loading video player...
Prabhat Sharma, founder and CEO of OpenObserve, explains how their AI-native open source observability platform delivers performance improvements while dramatically reducing costs compared to legacy solutions like DataDog, Elasticsearch, and Splunk. OpenObserve processes over 2.5 petabytes of data per day for their largest customer while providing 10x better analytics performance and up to 140x lower storage costs. Built with Rust and Parquet, the platform unifies logs, metrics, traces, and LLM observability in a single solution that runs on a fraction of the infrastructure required by traditional tools. The discussion covers technical architecture decisions, handling hyperscaler limitations, AI-powered SRE agents, and the vision for UI-less observability where AI automatically detects and remediates issues without human intervention. Key takeaways: • OpenObserve reduces infrastructure from 5 nodes to 1 while maintaining performance • Storage costs can be 140x lower than Elasticsearch with better compression • Enterprise features are free for companies ingesting under 200GB per day • AI SRE agent correlates logs and automates problem detection and remediation • Built on Rust and Parquet for extreme speed and efficiency • Supports OpenTelemetry, Telegraph, Syslog for comprehensive data collection Chapters: 0:15 - Introduction to OpenObserve 1:42 - OpenObserve vs legacy observability solutions 2:43 - Performance improvements and cost savings 3:51 - Running on hyperscalers and hitting Google Cloud limits 7:37 - Technology stack and architecture decisions 18:04 - AI SRE agent and LLM observability 19:32 - The future of UI-less observability Relevant keywords: OpenObserve, observability platform, open source monitoring, DataDog alternative, Elasticsearch alternative, Splunk alternative, LLM observability, AI SRE agent, Rust programming, Parquet storage, OpenTelemetry, cloud native, Kubernetes observability, log management, metrics monitoring, distributed tracing