Laurent Goujon

People often ask me about the differences in architecture between Amazon Dynamo (the 2007 SOSP paper), DynamoDB (the AWS serverless NoSQL database), and Aurora DSQL (the AWS serverless SQL databases). I captured the key points on durability, high availability, programming model, and trade-offs in my blog: https://lnkd.in/gkMG_tVD TL;DR: Despite the shared name, Dynamo and DynamoDB have little in common architecturally, and make very different trade-off decisions.

631

8 Comments

"Build a Data Lakehouse with Apache Iceberg, Polaris, Trino & MinIO" Enjoyed reading this two-part blog post series by Gilles Philippart, teaching how to run a Data Lakehouse locally, using open-source technologies: - MinIO for object storage - Apache Iceberg as the open table format - Apache Polaris as the metadata catalog - Trino as the query engine. The second part adds Apache Kafka and Flink for streaming data ingest. I also learned about JR, a CLI utility for creating randomized Kafka test data. Everything running locally on your machine via Docker Compose, which is great for learning and experimenting. Really neat, definitely recommend to check it out! ���� https://lnkd.in/e73fSnPS

201

13 Comments

we just open-sourced the SigNoz MCP Server — let Claude, Cursor, and your agents talk to your observability stack 🚀 AI assistants are great at reasoning, but they’re blind without context. With the SigNoz MCP Server, your assistant can securely fetch the right observability data on demand — inside your IDE or coding assistant — to debug faster, review incidents, and answer on-call questions in plain English. Some sample questions you can answer now and give context to your AI agent about: “What services were active in the last 24h?” “Which alert rules are firing right now?” “Show me the dashboard related to kafka.” but this is just the tip of the iceberg, you essentially have all your observability data available through natural language query now 📖 Why open source? SigNoz is built on open standards (OpenTelemetry). Open sourcing our MCP server continues our commitment to open source & open standards. 🙏 Try it / contribute - Works with SigNoz Cloud or self-hosted (grab an API key from Workspace Settings) - PRs welcome — especially new tools (logs/traces queries, runbook helpers, etc.) and example prompts Excited to see what our community starts building with it 🥳

190

24 Comments

After pioneering user-facing and agent-facing analytics, Apache Pinot and StarTree are now taking on a new frontier: powering analytical serving workloads directly from Apache Iceberg. Check out this excellent deep-dive from Neha Pawar and Gnanaguru Sattanathan: https://lnkd.in/gmhaEKXn It explains how Pinot achieves high-throughput, low-latency reads from Iceberg and how this architecture lets you tune for your workload to strike the right balance on the cost-performance spectrum. Proud of the team for continuing to innovate and push the boundaries of what’s possible in real-time analytics! 🚀

85

Excited to have two GAs for the Google Cloud Looker Platform live this week including: - period_over_period measures - Looker Enhanced System Activity Performance Views PoP is a new LookML measure type to make it easier for modelers to do PoP analysis in Looker without complex workarounds and data wrangling. And because it's in LookML it's governed and reusable. https://lnkd.in/eb2uFQh4 We've also launched into GA new enhanced Looker System Activity performance views (see gif and details in comments below). System Activity is an internal LookML model that connects to Looker's underlying application database to help Admins monitor user activity, content utilization, database performance and overall Looker app performance. https://lnkd.in/eb8KMfpD

128

10 Comments

You can now read and write to a Unity Catalog table from a Serverless EMR compute engine via the UC open API. And yes, it supports both Delta and Iceberg. Read more here- https://lnkd.in/eyc3VGCQ #unitycatalog #databricks

114

3 Comments

Business processes are becoming complex enough that you can't run your company on just SAP, Oracle, and MuleSoft. Companies are investing heavily in AI workflows, which must integrate with legacy systems, existing microservices, and external API's. Open-source stacks are increasingly replacing the monolithic, proprietary stacks that were popular 5-10 years ago (especially before Kubernetes). These trends are only going to accelerate—and it means that you need to have a software engineering organization and architecture that is: - open-source - cloud-capable and cloud-agnostic - built for the demands of the modern enterprise. Colt McNealy and I have been working on such a platform, based on the open-source LittleHorse Kernel (and the surrounding Saddle ecosystem) which I am very excited about. You can check out the core Kernel on GitHub, and get in touch with Colt on Slack if you want to learn more! https://lnkd.in/gsEDVgDR

83

5 Comments

IBM is back! They just joined Hugging Face Enterprise & released Granite 4.0 in open-source with a new hybrid Mamba/transformer architecture that reduces memory requirements without reducing accuracy much. This set of models are great for agentic workflows like tool calling, document analysis, RAG, especially in an enterprise setup 🚀 The "Micro" (3.4B) model can even run 100% locally in your browser on WebGPU, powered by 🤗 TransformersJS! 3B dense hybrid: https://lnkd.in/exabrV-t 3B MoE with 1B active: https://lnkd.in/e98GVG_5 32B MoE with 9B active: https://lnkd.in/eX4hUdVR 🗂️ Full Model collection: https://lnkd.in/efwqj939 🔗 In-browser demo: https://lnkd.in/ei_Pyqst

2,579

45 Comments