0 views collected around this technical thread.
This article explains the evolution and inner workings of Spark's shuffle phase, comparing the original Hash‑based shuffle, the default Sort‑based shuffle, the optimized Tungsten‑Sort shuffle, and related configuration options that affect performance and file handling in large‑scale data processing.
This article continues the deep dive into Apache Spark memory management, explaining storage memory handling—including RDD persistence, caching, eviction, and disk spilling—as well as execution memory allocation for multi-tasking and shuffle operations, and detailing Spark’s internal structures such as BlockManager, StorageLevel, and Tungsten page management.
The article explains how Spark 2.0’s second‑generation Tungsten engine replaces the traditional Volcano iterator model with whole‑stage code generation and vectorization, eliminating virtual calls, keeping temporary data in CPU registers, and using loop unrolling and SIMD to achieve order‑of‑magnitude performance gains on large‑scale data workloads.
This article reviews Spark's evolution from version 1.2 to 1.6, explains the DataFrame and Tungsten projects, shares Hulu’s real‑world Spark deployments, and discusses performance‑related challenges such as stack overflow, streaming receiver latency, and class‑loader deadlocks.