I'm noticing a growing trend of using #S3 instead of traditional databases or block storage in products, which significantly reduces costs. We've transitioned to using #Quickwit, an analytics and search engine for #OpenTelemetry tracing and logging. It can also be used as a #Jaeger storage backend. You can easily integrate Jaeger with #Grafana, using Quickwit for S3-based storage. All data is stored in S3, resulting in a time-to-search of just 45 seconds, which is impressive. The time-to-read is nearly 70 ms. Quickwit is written in Rust and packaged as a single binary, easy to install and deploy. Combined with Grafana, Jaeger and S3, it creates a powerful setup for observability. Quickwit stores logs and traces in S3, while Grafana provides a flexible dashboard for visualizing and analyzing the data. This combination offers fast search performance and scalable storage without the complexity of traditional ELK stacks—no Java, yay! #Turbopuffer also caught my attention—a serverless vector database that leverages S3 for storage. It offers a cost advantage of 10x-100x compared to alternative vector database solutions. The biggest user of this technology is Cursor.ai, which is my favorite AI tool for daily use.