Distributed Stream Processing Systems (DSPSs) form the backbone of real-time processing and analytics at ByteDance, where Apache Flink powers one of the largest production clusters worldwide. Ensuring resiliency, the ability to withstand and rapidly recover from failures, together with operational stability, which provides consistent and predictable performance under normal conditions, is essential for meeting strict Service Level Objectives (SLOs). However, achieving resiliency and stability in large-scale production environments remains challenging due to the cluster scale, business diversity, and significant operational overhead. In this work, we present StreamShield, a production-proven resiliency solution deployed in ByteDance's Flink clusters. Designed along complementary perspectives of the engine and cluster, StreamShield introduces key techniques to enhance resiliency, covering runtime optimization, fine-grained fault-tolerance, hybrid replication strategy, and high availability under external systems. Furthermore, StreamShield proposes a robust testing and deployment pipeline that ensures reliability and robustness in production releases. Extensive evaluations on a produc
With the rapid rise of intelligent data services, modern enterprises increasingly require efficient, multimodal, and cost-effective data analytics infrastructures. However, in ByteDance's production environments, existing systems fall short due to limitations such as I/O-inefficient multimodal storage, inflexible query optimization (e.g., failing to optimize multimodal access patterns), and performance degradation caused by resource disaggregation (e.g., loss of data locality in remote storage). To address these challenges, we introduce ByteHouse (https://bytehouse.cloud), a cloud-native data warehouse designed for real-time multimodal data analytics. The storage layer integrates a unified table engine that provides a two-tier logical abstraction and physically consistent layout, SSD-backed cluster-scale cache (CrossCache) that supports shared caching across compute nodes, and virtual file system (NexusFS) that enable efficient local access on compute nodes. The compute layer supports analytical, batch, and incremental execution modes, with tailored optimizations for hybrid queries (e.g., runtime filtering over tiered vector indexes). The control layer coordinates global metadata a
Cardinality estimation is a critical component and a longstanding challenge in modern data warehouses. ByteHouse, ByteDance's cloud-native engine for extensive data analysis in exabyte-scale environments, serves numerous internal decision-making business scenarios. With the increasing demand for ByteHouse, cardinality estimation becomes the bottleneck for efficiently processing queries. Specifically, the existing query optimizer of ByteHouse uses the traditional Selinger-like cardinality estimator, which can produce substantial estimation errors, resulting in suboptimal query plans. To improve cardinality estimation accuracy while maintaining a practical inference overhead, we develop a framework ByteCard that enables efficient training and integration of learned cardinality estimators. Furthermore, ByteCard adapts recent advances in cardinality estimation to build models that can balance accuracy and practicality (e.g., inference latency, model size, training overhead). We observe significant query processing speed-up in ByteHouse after replacing the existing cardinality estimator with ByteCard for several optimization scenarios. Evaluations on real-world datasets show the integra
The training scale of large language models (LLMs) has reached tens of thousands of GPUs and is still continuously expanding, enabling faster learning of larger models. Accompanying the expansion of the resource scale is the prevalence of failures (CUDA error, NaN values, job hang, etc.), which poses significant challenges to training stability. Any large-scale LLM training infrastructure should strive for minimal training interruption, efficient fault diagnosis, and effective failure tolerance to enable highly efficient continuous training. This paper presents ByteRobust, a large-scale GPU infrastructure management system tailored for robust and stable training of LLMs. It exploits the uniqueness of LLM training process and gives top priorities to detecting and recovering failures in a routine manner. Leveraging parallelisms and characteristics of LLM training, ByteRobust enables high-capacity fault tolerance, prompt fault demarcation, and localization with an effective data-driven approach, comprehensively ensuring continuous and efficient training of LLM tasks. ByteRobust is deployed on a production GPU platform and achieves 97% ETTR for a three-month training job on 9,600 GPUs.
While scaling laws for recommendation models have gained significant traction, existing architectures such as Wukong, HiFormer and DHEN, often struggle with sub-optimal designs and hardware under-utilization, limiting their practical scalability. Our previous TokenMixer architecture (introduced in RankMixer paper) addressed effectiveness and efficiency by replacing self-attention with a ightweight token-mixing operator; however, it faced critical bottlenecks in deeper configurations, including sub-optimal residual paths, vanishing gradients, incomplete MoE sparsification and constrained scalability. In this paper, we propose TokenMixer-Large, a systematically evolved architecture designed for extreme-scale recommendation. By introducing a mixing-and-reverting operation, inter-layer residuals and the auxiliary loss, we ensure stable gradient propagation even as model depth increases. Furthermore, we incorporate a Sparse Per-token MoE to enable efficient parameter expansion. TokenMixer-Large successfully scales its parameters to 7-billion and 15-billion on online traffic and offline experiments, respectively. Currently deployed in multiple scenarios at ByteDance, TokenMixer-Large has
Modern microservice systems exhibit continuous structural evolution in their runtime call graphs due to workload fluctuations, fault responses, and deployment activities. Despite this complexity, our analysis of over 500,000 production traces from ByteDance reveals a latent regularity: execution paths concentrate around a small set of recurring invocation patterns. However, existing resource management approaches fail to exploit this structure. Industrial autoscalers like Kubernetes HPA ignore inter-service dependencies, while recent academic methods often assume static topologies, rendering them ineffective under dynamic execution contexts. In this work, we propose Morphis, a dependency-aware provisioning framework that unifies pattern-aware trace analysis with global optimization. It introduces structural fingerprinting that decomposes traces into a stable execution backbone and interpretable deviation subgraphs. Then, resource allocation is formulated as a constrained optimization problem over predicted pattern distributions, jointly minimizing aggregate CPU usage while satisfying end-to-end tail-latency SLOs. Our extensive evaluations on the TrainTicket benchmark demonstrate th
Recently, industrial pioneers like Amazon, Tencent, ByteDance, and Huawei have been adopting BBR as their congestion control algorithm for live-streaming applications, including TikTok Live. However, BBR, originally crafted for bulk data transmission, faces multiple challenges in live-streaming scenarios. In this paper, we first explore two key issues associated with BBR due to inaccurate bandwidth estimation in live-streaming scenarios: (i) BBR cannot easily exit its startup phase, resulting in a fierce self-inflicted loss. (ii) BBR sends data at a lower rate than the available bandwidth during its stable phase. We then propose BBR-Copilot, an auxiliary congestion control component that cooperates with BBR, making BBR better adapt to live-streaming scenarios. BBR-Copilot allows for proactively generating accurate bandwidth measurement samples by smartly creating and sending extra data. We implement the BBR-Copilot prototype upon QUIC and evaluate it via testbed. Experimental evaluation results show that BBR-Copilot effectively enhances BBR's performance in live-streaming scenarios.
Modern cloud applications often comprise thousands of microservices whose interactions form complex request paths. Traditional inter-service access control restricts individual service-to-service requests, but fails to prevent multi-hop attacks, where each hop appears legitimate yet the overall path violates security intent. This gap leaves systems exposed to unauthorized access and data exfiltration. In this paper, we present eMicro, a path-aware defense system for microservices that prevents such attacks while remaining efficient and deployable. eMicro enforces real-time multi-hop access control through three key techniques: (1) history-based access control extended to capture service invocation sequences; (2) security policies encoded as efficient deterministic finite automaton (DFA), supporting constant-time lookups and compact label propagation; (3) eBPF-based in-kernel request tracing for transparent, low-overhead enforcement without code changes. Evaluations on DeathStarBench and production cloud traces from Uber, Alibaba, and ByteDance, covering 12 million request workflows and thousands of services, demonstrate the scalability of eMicro. eMicro performs policy checks in 1
Driven by scaling laws, recommender systems increasingly rely on larger-scale models to capture complex feature interactions and user behaviors, but this trend also leads to prohibitive training and inference costs. While long-sequence models can reuse user-side computation through KV Caching, such reuse is difficult in TokenMixer-based dense feature interaction architectures, where user and group features are deeply entangled and mixed-up across layers. In this work, we present User-Group Separation (UG-Sep), an industrial large-scale framework that enables user-side computation reusable in TokenMixer-based dense interaction models for the first time. UG-Sep explicitly disentangles user-side and item-side information flows within token-mixing layers, ensuring that a subset of tokens preserves purely user-side representations across layers. This design allows the corresponding per-token computations to be reused across multiple samples, significantly reducing redundant inference cost. To compensate for the potential expressive capacity loss induced by masking, we further propose an Information Compensation strategy that adaptively reconstructs suppressed user-item interactions. Mor
We present Seed1.8, a foundation model aimed at generalized real-world agency: going beyond single-turn prediction to multi-turn interaction, tool use, and multi-step execution. Seed1.8 keeps strong LLM and vision-language performance while supporting a unified agentic interface-search, code generation and execution, and GUI interaction. For deployment, it offers latency- and cost-aware inference, including configurable thinking modes and optimized visual encoding for images and video. We report evaluations on standard benchmarks and application-aligned workflows spanning foundational skills, multimodal understanding, and agentic behavior. Seed1.8 is released to support further research and development on interactive, real-world use cases.
Large language model (LLM) training is one of the most demanding distributed computations today, often requiring thousands of GPUs with frequent synchronization across machines. Such a workload pattern makes it susceptible to stragglers, where the training can be stalled by few slow workers. At ByteDance we find stragglers are not trivially always caused by hardware failures, but can arise from multiple complex factors. This work aims to present a comprehensive study on the straggler issues in LLM training, using a five-month trace collected from our ByteDance LLM training cluster. The core methodology is what-if analysis that simulates the scenario without any stragglers and contrasts with the actual case. We use this method to study the following questions: (1) how often do stragglers affect training jobs, and what effect do they have on job performance; (2) do stragglers exhibit temporal or spatial patterns; and (3) what are the potential root causes for stragglers?
Modern REST API testing faces a critical challenge in defining reliable test oracles, particularly in agile industrial environments where formal specifications (e.g., OpenAPI) are frequently missing or outdated, and historical execution logs are unavailable for newly deployed endpoints. In this paper, we present Restor (Reinforcement Enhanced Single-Traffic Oracle generator for REST APIs), a framework that generates executable test assertions from a single observed request-response pair in a black-box setting. Unlike existing approaches that rely on rule-based templates or massive training logs, Restor utilizes a novel data augmentation pipeline to fine-tune a lightweight Large Language Model (LLM) via Group Relative Policy Optimization (GRPO). This training process enables the model to internalize testing "common sense" by optimizing a reward function that jointly encourages: (i) the selection of stable, semantically meaningful fields for validation and the avoidance of dynamic noise (e.g., timestamps or trace IDs); (ii) the generation of robust assertions that withstand logic variations. We evaluate Restor on an industrial dataset comprising over 2,300 API traces across 246 real-
Large foundation models are integrated into Computer Use Agents (CUAs), enabling autonomous interaction with operating systems through graphical user interfaces (GUIs) to perform complex tasks. This autonomy introduces serious security risks: malicious instructions or visual prompt injections can trigger unsafe reasoning and cause harmful system-level actions. Existing defenses, such as detection-based blocking, prevent damage but often abort tasks prematurely, reducing agent utility. In this paper, we present MirrorGuard, a plug-and-play defense framework that uses simulation-based training to improve CUA security in the real world. To reduce the cost of large-scale training in operating systems, we propose a novel neural-symbolic simulation pipeline, which generates realistic, high-risk GUI interaction trajectories entirely in a text-based simulated environment, which captures unsafe reasoning patterns and potential system hazards without executing real operations. In the simulation environment, MirrorGuard learns to intercept and rectify insecure reasoning chains of CUAs before they produce and execute unsafe actions. In real-world testing, extensive evaluations across diverse b
Speech and audio generation is often needed in animation dubbing, audio drama, movies, advertising, games, podcasts, and short-video production. In these scenarios, creators may need to design voices without reference recordings, control speaker styles with natural language, support acoustic scenes with environments and audio effects, and later reuse the designed voices. Therefore, it is important to support multi-speaker speech and audio generation for both instruct and zero-shot tasks. The instruct task requires a caption of the environment, speaker styles, and fine-grained content, while the zero-shot task uses reference audio together with the same fine-grained content. We address these tasks from both the data and model sides. First, we propose SwanData-Caption, which cleans raw speech and audio data, adds targeted synthetic coverage, and annotates diverse and accurate multi-level captions. Then, we propose SwanTale, a multi-speaker expressive speech and audio generation model that supports both zero-shot and instruct tasks. We introduce SwanVAE to support high-quality multi-audio-modality generation. Then, we adopt reward-conditioned quality control and Engram conditioning, a
A B testing serves as the gold standard for large scale, data driven decision making in online businesses. To mitigate metric variability and enhance testing sensitivity, control variates and regression adjustment have emerged as prominent variance reduction techniques, leveraging pre experiment data to improve estimator performance. Over the past decade, these methods have spawned numerous derivatives, yet their theoretical connections and comparative properties remain underexplored. In this paper, we conduct a comprehensive analysis of their statistical properties, establish a formal bridge between the two frameworks in practical implementations, and extend the investigation from design based to model-based frameworks. Through simulation studies and real world experiments at ByteDance, we validate our theoretical insights across both frameworks. Our work aims to provide rigorous guidance for practitioners in online controlled experiments, addressing critical considerations of internal and external validity. The recommended method control variates with group specific coefficient estimates has been fully implemented and deployed on ByteDance's experimental platform.
Code review remains a critical yet resource-intensive process in software development, particularly challenging in large-scale industrial environments. While Large Language Models (LLMs) show promise for automating code review, existing solutions face significant limitations in precision and practicality. This paper presents BitsAI-CR, an innovative framework that enhances code review through a two-stage approach combining RuleChecker for initial issue detection and ReviewFilter for precision verification. The system is built upon a comprehensive taxonomy of review rules and implements a data flywheel mechanism that enables continuous performance improvement through structured feedback and evaluation metrics. Our approach introduces an Outdated Rate metric that can reflect developers' actual adoption of review comments, enabling automated evaluation and systematic optimization at scale. Empirical evaluation demonstrates BitsAI-CR's effectiveness, achieving 75.0% precision in review comment generation. For the Go language which has predominant usage at ByteDance, we maintain an Outdated Rate of 26.7%. The system has been successfully deployed at ByteDance, serving over 12,000 Weekly
Video generation is rapidly evolving towards unified audio-video generation. In this paper, we present ALIVE, a generation model that adapts a pretrained Text-to-Video (T2V) model to Sora-style audio-video generation and animation. In particular, the model unlocks the Text-to-Video&Audio (T2VA) and Reference-to-Video&Audio (animation) capabilities compared to the T2V foundation models. To support the audio-visual synchronization and reference animation, we augment the popular MMDiT architecture with a joint audio-video branch which includes TA-CrossAttn for temporally-aligned cross-modal fusion and UniTemp-RoPE for precise audio-visual alignment. Meanwhile, a comprehensive data pipeline consisting of audio-video captioning, quality control, etc., is carefully designed to collect high-quality finetuning data. Additionally, we introduce a new benchmark to perform a comprehensive model test and comparison. After continue pretraining and finetuning on million-level high-quality data, ALIVE demonstrates outstanding performance, consistently outperforming open-source models and matching or surpassing state-of-the-art commercial solutions. With detailed recipes and benchmarks, we
While diffusion models dominate the field of visual generation, they are computationally inefficient, applying a uniform computational effort regardless of different complexity. In contrast, autoregressive (AR) models are inherently complexity-aware, as evidenced by their variable likelihoods, but are often hindered by lossy discrete tokenization and error accumulation. In this work, we introduce Generative Refinement Networks (GRN), a next-generation visual synthesis paradigm that addresses these issues. At its core, GRN addresses the discrete tokenization bottleneck through a theoretically near-lossless Hierarchical Binary Quantization (HBQ), achieving a reconstruction quality comparable to continuous counterparts. Built upon HBQ's latent space, GRN fundamentally upgrades AR generation with a global refinement mechanism that progressively perfects and corrects artworks -- like a human artist painting. Besides, GRN integrates an entropy-guided sampling strategy, enabling complexity-aware, adaptive-step generation without compromising visual quality. On the ImageNet benchmark, GRN establishes new records in image reconstruction (0.56 rFID) and class-conditional image generation (1.
As the complexity of mobile applications grows exponentially and the fragmentation of user device environments intensifies, ensuring online application stability faces unprecedented challenges. Traditional methods, such as static logging and post-crash analysis, lack real-time contextual information, rendering them ineffective against "ghost bugs" that only manifest in specific scenarios. This highlights an urgent need for dynamic runtime observability: intercepting and tracing arbitrary methods in production without requiring an app release. We propose XTrace, a novel dynamic tracing framework. XTrace introduces a new paradigm of non-invasive proxying, which avoids direct modification of the virtual machine's underlying data structures. It achieves high-performance method interception by leveraging and optimizing the highly stable, built-in instrumentation mechanism of the Android ART virtual machine. Evaluated in a ByteDance application with hundreds of millions of daily active users, XTrace demonstrated production-grade stability and performance. Large-scale online A/B experiments confirmed its stability, showing no statistically significant impact (p > 0.05) on Crash User Ra
Multi-tenant architectures enhance the elasticity and resource utilization of NoSQL databases by allowing multiple tenants to co-locate and share resources. However, in large-scale cloud environments, the diverse and dynamic nature of workloads poses significant challenges for multi-tenant NoSQL databases. Based on our practical observations, we have identified three crucial challenges: (1) the impact of caching on performance isolation, as cache hits alter request execution and resource consumption, leading to inaccurate traffic control; (2) the dynamic changes in traffic, with changes in tenant traffic trends causing throttling or resource wastage, and changes in access distribution causing hot key pressure or cache hit ratio drops; and (3) the imbalanced layout of data nodes due to tenants' diverse resource requirements, leading to low resource utilization. To address these challenges, we introduce ABase, a multi-tenant NoSQL serverless database developed at ByteDance. ABase introduces a two-layer caching mechanism with a cache-aware isolation mechanism to ensure accurate resource consumption estimates. Furthermore, ABase employs a predictive autoscaling policy to dynamically ad