Big Data Analytics: Distributed Cluster Simulator

Big data systems like Hadoop and Spark turn an incoming stream of records into blocks, spread those blocks across a cluster of worker nodes by hashing a partition key, and replicate every block onto extra nodes so a single disk or machine failure never loses data. This simulator renders that pipeline directly: watch data blocks leave the ingest stream, land on a hash-partitioned primary node (blue), and fan out to replica nodes (amber) for fault tolerance, while each node's height and color track how close it is to saturation. Cluster size, ingestion rate, replication factor and the batch-vs-streaming engine are all live controls, and the throughput/utilization/latency readouts are driven by a real M/M/1 queueing model rather than a canned animation.