In the AI era, we talk a lot about models, agents, and intelligence. But underneath all of it, data still has to move, persist, replicate, and recover, fast and predictably. Apache Iggy 0.9.0 is a major step forward. 🚀 Clustering is here, built around Viewstamped Replication Revisited (VSR). And we’re doing it while staying focused on the thing Iggy has cared about from day one: predictable tail latency. On a 3-node AWS i4i cluster at 800 MB/s: Producer, replicated: ~3 ms p99 Producer, persisted to disk: ~7 ms p99 Warm consumer: <1 ms p99 I’m incredibly proud of what the Apache Iggy community has built. More than 60 contributors and 600+ merged PRs went into this release, including 34 first-time contributors. Thank you to everyone who wrote code, reviewed PRs, tested releases, reported issues, challenged assumptions, and helped move the project forward. And we’re not settling here! There are already further performance optimizations underway for 0.9.1. The goal isn’t simply to make Iggy fast. It’s to keep asking how much latency, overhead, and infrastructure we can eliminate while preserving durability and correctness. Modern AI infrastructure deserves a streaming engine designed with modern systems engineering. Congratulations, @ApacheIggy community. Onward. 👏 #ApacheIggy #MessageStreaming #AIDataInfrastructure

So while @Apple starts its next chapter, we’re starting one of our own, for real-time streaming and agent data infrastructure. September 9 will be a special day for @laserdatainc too. One more stream is here. 💚 #LaserData #AIDataInfrastructure #DataStreaming #AGDXProtocol #LaserSDK