DoorDash has developed Entity Cache, a transparent proxy caching platform built on Envoy and Valkey to reduce redundant service-to-service requests across its microservices architecture. Operating...

钉住官方 File Locking And Concurrency 文档定义的 UNLOCKED/SHARED/RESERVED/PENDING/EXCLUSIVE 五态锁阶梯：谁能与谁共存、PENDING 如何防写者饿死、Pager 为何只跟踪四态；用本机 3.53.2 真实双连接冲突实验验证，shared cache 与 WAL 差异只点边界，留给第 8、10 篇。

拆解 Pager 作为 B-Tree 与操作系统文件之间的契约层：page cache 命中/未命中路径、脏页生命周期与提交前的可回滚保证，并与 PG shared_buffers、InnoDB Buffer Pool 的跨进程共享模型对照。

本文讨论了RocksDB的读取性能优化，重点在于Table Cache和Block Cache的作用。Table Cache缓存已打开的TableReader，避免重复打开SST文件；Block Cache缓存数据块，减少I/O开销。使用Bloom Filter和Ribbon Filter可以跳过不存在的键的I/O，Partitioned Index则通过分割大型索引降低读取成本。调参时需综合考虑文件数和缓存策略，以优化读取性能。

算子库工程涉及多个kernel的选择与优化，主要包括dispatch和autotune机制。dispatch根据形状、精度和架构选择最佳kernel，autotune通过实测候选配置优化性能。JIT编译按需生成特定kernel，提高效率，但需缓存编译结果以减少延迟。此外，算子库需关注正确性回归和版本管理，以确保稳定性和可维护性。

effective_cache_size doesn't allocate memory, reserve RAM, or control runtime behavior—it merely whispers a number to the query planner to make it smarter…

Two recent Linux kernel vulnerabilities have been disclosed: Copy Fail (CVE-2026-31431) on April 29, 2026, and Dirty Frag (CVE-2026-43284 and CVE-2026-43500) on May 7, 2026. Both allow local users...

Netflix improves Apache Druid performance with interval aware caching, serving 84% of analytics results from cache and reducing query load by 33%. The system decomposes rolling window queries into...

Cloudflare recently introduced its Gen 13 servers, marking a shift in how its network handles traffic. Instead of relying on large CPU caches for speed, the company redesigned its software to...

Cloudflare and ETH Zurich highlight how AI-driven crawler traffic challenges traditional caching in CDNs and databases. They propose AI-aware strategies including separate cache tiers, adaptive...

Claude Code的fork子agent通过共享缓存机制优化并行任务成本，确保99%以上的字节一致性，显著降低输入费用。使用Anthropic API的Prompt Cache机制，缓存命中可享受90%折扣，减少并行调用成本。设计强调字节级一致性，确保多个请求共享相同前缀，避免不必要费用。

PostgreSQL的查询规划器在执行预处理语句时，前五次使用自定义计划，第六次可能切换到通用计划。这种切换可能导致性能下降，特别是在数据分布不均时。用户应注意这一行为，并在必要时强制使用自定义计划以优化性能。