# concurrency control (blogs) — RSS Amplifier Recent posts from the 2 feeds in the RSS Amplifier directory that cover concurrency control. Page: Feed: --- ## [Notes on CRAQ](https://flaneur2020.github.io/posts/2026-05-24-craq/) _2026-05-24 · Flaneur2020_ Recently while studying 3fs, I learned that it uses CRAQ (Chain Replication with Apportioned Queries) as its replication algorithm. Coincidentally, I had similar requirements at work, so I’m documenting my understanding of CRAQ here. CRAQ is already a quite popular engineering solution for replication in object storage or KV systems. CRAQ can be viewed as a simple enhancement to traditional… ## [AI and the Curse of Claudia](https://shivamkapoor.com/2026/04/04/ai-and-the-curse-of-claudia/) _2026-04-04 · Shivam Kapoor · Thinking Out Loud_ “Our bodies will stay always as they are, fresh and alive, but we must never hesitate to bring death, because it is how we live.” — Lestat, Interview with the Vampire The unwritten rule among vampires was broken the night Claudia was made. She was never supposed to ## [Why I Built a Custom Electron App Just to Use iCloud Mail](https://shivamkapoor.com/2026/03/14/why-i-built-a-small-electron-app-just-to-use-icloud-mail/) _2026-03-14 · Shivam Kapoor · Thinking Out Loud_ I run dark mode everywhere macOS, browser, editor, terminal. At this point its not really a preference anymore. Bright white screens for hours genuinely strain my eyes. That setup worked perfectly until I started using Hide My Email, Apple’s privacy feature that generates random relay addresses for signing ## [A Walkthrough of nano-vllm](https://flaneur2020.github.io/posts/2025-10-12-nano-vllm/) _2025-10-12 · Flaneur2020_ Recently, I’ve been delving into the architecture of production-grade inference engines. While projects like vLLM and SGLang are crazy sophisticated, their complexity can make detailed code tracing difficult. Over the holidays, I came across nano-vllm—a tightly scoped codebase that still delivers end-to-end support for Page Attention and scheduling. I took a deep dive into its internals and… ## [Why Bloom Filters Swipe Right on Immutability](https://shivamkapoor.com/2025/09/04/why-bloom-filters-swipe-right-on-immutability/) _2025-09-04 · Shivam Kapoor · Thinking Out Loud_ In my last post Bloom Filters: MathMania , we worked through the derivations for optimal number of bits to allocate and the right number of hash functions to use – striking a balance between space, computation, and false positives. But all of that math rested on one simple assumption: the set ## [Notes on RL: Policy Gradient & Log Derivative Trick](https://flaneur2020.github.io/posts/2025-02-25-policy-gradient/) _2025-02-25 · Flaneur2020_ Recently I followed this tutorial to implement a basic Q-Learning reinforcement learning algorithm for Flappy Bird, which surprisingly worked quite well, reaching over 10,000 points. Q-Learning essentially estimates a value $Q(s, a)$ for each state-action pair $(s, a)$, and then selects actions based on these $Q$ values. Deep Q-Network (DQN) builds on basic Q-Learning by adding a neural network to… ## [Bloom Filters Mathmania](https://shivamkapoor.com/2025/01/13/bloom-filters-mathmania/) _2025-01-13 · Shivam Kapoor · Thinking Out Loud_ So, you’ve come across Bloom filters and understand that, despite their probabilistic nature, they are a great fit for your use case. You’ve decided to integrate them into your system design, but you’re unsure about the optimal size and the number of hash functions ## [Using Binomial Distribution to Model Data Durability](https://shivamkapoor.com/2024/11/14/using-binomial-distribution-to-model-data-durability/) _2024-11-14 · Shivam Kapoor · Thinking Out Loud_ Durability requirements influence the choice of data protection mechanisms, such as replication, erasure coding, and RAID parity configurations. Achieving higher durability involves trade-offs between redundancy , storage usage ratio , and computational complexity . Replication achieves durability by creating multiple copies of data, which increases redundancy but reduces the storage usage ratio.…