Tagged articles
batching
19 articles · Page 1 of 1
This article explains why GPUs underutilize during LLM inference and compares static, dynamic, and continuous batching strategies, detailing their trade-offs, the role of KV cache, iteration-level scheduling, and chunked prefill to optimize throughput for variable-length outputs.
GPU utilizationKV cacheLLM inference
0 likes · 13 min read

Kafka’s exceptionally high throughput stems from a combination of design choices—including sequential disk writes, zero‑copy data transfer, batch processing, partition‑level parallelism, OS page‑cache utilization, and producer‑side compression—that together minimize I/O overhead, CPU usage, and latency.
CompressionKafkaPage Cache
0 likes · 6 min read

The article analyses how synchronous Kafka producer calls block business threads on ACK, limiting throughput, and shows that moving to asynchronous sending with the Accumulator decouples submission from confirmation, unlocking batch, compression, and back‑pressure mechanisms that enable stable operation at tens of millions of QPS.
Kafkaasynchronousbackpressure
0 likes · 21 min read

This article presents a production‑grade Flink ClickHouse sink that solves common pain points such as lack of size‑based batching, static table schemas, and distributed‑table latency by introducing data‑size batching, dynamic table routing, local‑table writes, load‑balanced node discovery, back‑pressure queues, dual‑trigger flush, and recursive retry with node exclusion, all integrated with Flink checkpoint semantics for at‑least‑once guarantees.
ClickHouseDynamic ShardingFlink
0 likes · 25 min read

This guide explains how to dramatically increase Kafka producer throughput by using batch sending, selecting appropriate compression algorithms, fine‑tuning producer configurations, employing asynchronous and concurrent sends, and optimizing client placement and network settings.
CompressionKafkabatching
0 likes · 4 min read

Facing 53.64 ms per‑image latency in a Flask‑served Vision Transformer classifier, we iteratively optimized the pipeline—switching to ONNX Runtime, leveraging TensorRT, replacing Pillow with OpenCV, eliminating URL downloads, and finally batching requests—reducing average server‑side processing to 8.34 ms, a 6.4× speedup.
FlaskONNXPython
0 likes · 28 min read

The article explains how Kafka achieves high throughput by writing messages sequentially to disk, leveraging OS page cache and zero‑copy system calls, using partitioned topics for parallelism, batching and compressing records on both producer and broker sides, and employing asynchronous replication with configurable persistence strategies.
CompressionKafkaPartitioning
0 likes · 4 min read

This guide explains how to unleash vLLM’s full potential by optimizing batch size, leveraging 4‑bit quantization, tuning concurrency parameters, planning capacity with token‑per‑second metrics, and implementing robust monitoring to balance latency, cost, and scalability in production deployments.
LLM servingPerformance Tuningbatching
0 likes · 10 min read

This article explains the core Kafka techniques—sequential disk writes, Linux page cache usage, producer/broker batching with compression, and zero‑copy data transfer via sendfile—that together enable massive concurrency and near‑memory write performance in large‑scale streaming architectures.
KafkaPage Cachebatching
0 likes · 4 min read

The article explains how Kafka attains high‑throughput performance by using sequential disk writes, leveraging the OS page cache, employing producer and consumer batching with configurable parameters, and utilizing zero‑copy sendfile to minimize CPU and memory overhead, enabling stable million‑message per second rates.
KafkaPage CacheSequential Write
0 likes · 5 min read

This guide explains how to efficiently produce messages with TDMQ CKafka, covering topic creation, partition sizing, retry policies, asynchronous sending, key/value handling, batch tuning, sticky partitioning, idempotence, ACK settings, code examples, and strategies to avoid data skew and ensure high‑throughput, low‑latency messaging.
CKafkaIdempotenceKafka
0 likes · 18 min read

This article explains how Kafka sustains massive traffic by writing logs sequentially to disk, leveraging the operating system’s page cache for fast in‑memory writes, employing zero‑copy techniques like sendfile to avoid user‑space copying, and batching messages to reduce network overhead, thereby delivering high‑throughput, low‑latency streaming.
KafkaPage Cachebatching
0 likes · 4 min read


JavaEdgeJun 27, 2025 · Artificial Intelligence
The article explains what inference engines are, why they are needed beyond raw Python scripts, and outlines best practices such as model quantization, batching, and parallelism, while comparing popular open‑source and commercial options for production AI workloads.
AI DeploymentLLMParallelism
0 likes · 14 min read

This article details practical techniques for boosting Kafka producer throughput, covering batch sending optimization, linger.ms tuning, asynchronous send handling, compression options, and concurrent sending strategies, while providing code examples and performance considerations for high‑throughput streaming applications.
CompressionKafkaasynchronous
0 likes · 4 min read

Kafka attains million‑level message throughput by writing logs sequentially to disk, leveraging OS page cache for batch flushing, sending messages in large batches, and employing zero‑copy techniques that move data directly from disk to network sockets, dramatically reducing I/O overhead and CPU usage.
KafkaPage Cachebatching
0 likes · 4 min read


ITPUBDec 10, 2022 · Backend Development
This article explores Kafka's performance secrets by examining network, disk, and algorithmic factors, detailing sequential writes, zero‑copy techniques, page‑cache usage, Reactor‑based networking, batching, compression, partition concurrency, and file‑structure optimizations, and provides practical guidance for high‑throughput deployments.
Reactor Modelbatchingdisk I/O
0 likes · 18 min read

This article explains the mechanics behind React re‑rendering, how state and props changes trigger renders, common pitfalls that cause multiple renders in class and function components, and practical strategies—including batching, state merging, useReducer, and refs—to minimize unnecessary updates.
Reactbatchingfrontend
0 likes · 22 min read

This article explains the GraphQL N+1 performance issue, demonstrates how DataLoader batches and caches database calls to eliminate redundant queries, walks through its core TypeScript implementation—including batch scheduling, cache handling, and load methods—while providing practical examples and integration tips for real‑world GraphQL servers.
CachingDataLoaderGraphQL
0 likes · 28 min read

This article explains the fundamentals and deep internals of React’s setState method, covering basic usage, asynchronous behavior, state‑update pitfalls, functional updates, and the underlying mechanisms such as enqueueSetState, enqueueUpdate, batchingStrategy, and transaction processing, illustrated with code snippets and diagrams.
ReactState Managementasynchronous updates
0 likes · 7 min read
