arXiv subjects

Zaoxing Liu

Publications and source records attributed to Zaoxing Liu.

At least 19 recordsLinked to original sources

arXiv2026

ASAP: Reimagining the Data Lifecycle using Application Semantic-Aware Processing

Across many domains (e.g., observability, networking, security), data processing pipelines face what we refer to as the CSP problem: achieving low Cost at large Scale, while maintaining high Performance. In response, we see several efforts to tackle CSP in various stages of the Collect-Transmit-Store-Analyze data lifecycle; such as approximate query processing in databases or sketches in network routers. Our work is driven by the simple insight: "seeing the forest for the trees". These proposed solutions (e.g., AQP, sketching, compression, rollups) addressing CSP share a common property - they exploit semantic-preserving opportunities to support application needs. In this paper, we make a case for ASAP, a paradigm that makes Application Semantic-Aware Processing (ASAP) a first-class design principle in data processing pipelines. We argue that by taking a unified view across ASAP primitives developed in different domains, across the entire data lifecycle, we can unlock new opportunities to tackle the CSP problem. In particular, we can: (i) enable novel cross-lifecycle optimizations such as analytics run directly on sketches computed at the source; (ii) leverage primitives developed in other application domains; and (iii) enable widespread adoption of these powerful techniques. We discuss research challenges in socializing the benefits of the ASAP paradigm, and show preliminary evidence that adopting ASAP can yield up to 3 orders of magnitude improvements in the CSP tradeoff for many application domains.

cs.DB↗

arXiv2026

AIChilles: Automatically Uncovering Hidden Weaknesses in AI-Evolved Systems

The computer systems community has recently seen growing interest in AI-driven system evolution, where AI agents iteratively rewrite systems. Frameworks such as AdaEvolve and Engram report 12-60% score improvements over human-designed algorithms. While these results are promising, there are practical concerns if these AI-evolved programs can perform worse on unseen workloads and exhibit scalability regressions. Given the speed and scale of AI-generated code, we need automated mechanisms to uncover such identify hidden weaknesses in AI-evolved systems programs. To this end, we develop AIChilles that takes as input a baseline program $P$ and an AI-evolved program $P'$, AIChilles searches for valid workloads where $P'$ regresses relative to $P$ in correctness, runtime, memory usage, or output quality. To tackle the diversity in system applications, weakness types and potential bugs, AIChilles combines deterministic workload-parameter extraction, agent-based constraint inference, differential oracles, and code-frequency coverage to discover diverse failures. Across five system applications and 30 AI-evolved programs, AIChilles finds 49 distinct hidden weaknesses. We also show that explicitly including AIChilles in the AI-driven development lifecycle can mitigate several of these weaknesses.

cs.AI↗

arXiv2026

Don't Let a Few Network Failures Slow the Entire AllReduce

Network failures are among the most frequent hardware faults in large-scale GPU clusters and a leading cause of training-job interruptions. Modern collective communication libraries such as NCCL mitigate network failures by rerouting traffic through surviving NICs on the same server, trading reduced inter-node bandwidth for uninterrupted training. However, the degraded server remains on the critical path of the standard ring algorithm, slowing the entire collective. We present the first information-theoretic lower bound on AllReduce completion time under asymmetric network bandwidth and show that when the straggler retains at least half of its original bandwidth, the unavoidable overhead relative to the fault-free optimum is only O(1/p) for p GPUs. We then design OptCC, a four-stage pipelined AllReduce algorithm that approaches this lower bound. Experiments on SimAI confirm that OptCC closes the gap left by existing fault-tolerant schemes: under practical network failures with up to 50% bandwidth loss, OptCC completes AllReduce within 2-6% of NCCL's fault-free ring performance, whereas the state-of-the-art incurs up to 57% overhead.

cs.DC↗

arXiv2026

Enabling Performant and Flexible Model-Internal Observability for LLM Inference

Today's inference-time workloads increasingly depend on timely access to a model's internal states. We present DMI-Lib, a high-speed deep model inspector that treats internal observability as a first-class systems primitive, decoupling it from the inference hot path via an asynchronous observability substrate built from Ring^2, a GPU-CPU memory abstraction for capturing and staging tensors, and a policy-controlled host backend that exports them. DMI-Lib enables the placement of observation points across a rich space of internal signals and diverse inference backends while preserving serving optimizations and adhering to tight GPU memory budgets. Our experiments demonstrate that DMI-Lib incurs only 0.4%--6.8% overhead in offline batch inference and an average of 6% in moderate online serving, reducing latency overhead by 2x-15x compared to existing baselines with similar observability features. DMI-Lib is open-sourced at https://github.com/ProjectDMX/DMI.

cs.LG↗

arXiv2026

Continuous Discovery of Vulnerabilities in LLM Serving Systems with Fuzzing

LLM inference and serving systems have become security-critical infrastructure; however, many of their most concerning failures arise from the serving layer rather than from model behavior alone. Modern inference engines combine KV cache, batching, prefix sharing, speculative decoding, adapters, and multi-tenant scheduling, creating shared-state behavior that only emerges under realistic concurrent workloads and is missed by standard model, safety, and API tests. We present GRIEF, a greybox fuzzer for LLM inference engines that treats timed multi-request traces as first-class inputs, uses lightweight oracles to detect crashes, hangs, performance pathologies, and silent output corruption, and applies controlled replay with log-probability checks to confirm reproducible serving-layer failures. Across early campaigns on vLLM and SGLang, GRIEF discovers 15 vulnerabilities, 10 confirmed by engine developers, including 2 CVEs, spanning KV-cache isolation failures, cross-request performance interference, and crash or liveness bugs. These results show that concurrency, caching, and state reuse can induce silent cross-request contamination, noisy-neighbor denial of service, and delayed crashes without malformed inputs or explicit server errors, making concurrent serving behavior a first-class security and reliability boundary for LLM infrastructure.

cs.CR↗

arXiv2025

Reliable and Resilient Collective Communication Library for LLM Training and Serving

Modern ML training and inference now span tens to tens of thousands of GPUs, where network faults can waste 10--15\% of GPU hours due to slow recovery. Common network errors and link fluctuations trigger timeouts that often terminate entire jobs, forcing expensive checkpoint rollback during training and request reprocessing during inference. We present R$^2$CCL, a fault-tolerant communication library that provides lossless, low-overhead failover by exploiting multi-NIC hardware. R$^2$CCL performs rapid connection migration, bandwidth-aware load redistribution, and resilient collective algorithms to maintain progress under failures. We evaluate R$^2$CCL on two 8-GPU H100 InfiniBand servers and via large-scale ML simulators modeling hundreds of GPUs with diverse failure patterns. Experiments show that R$^2$CCL is highly robust to NIC failures, incurring less than 1\% training and less than 3\% inference overheads. R$^2$CCL outperforms baselines AdapCC and DejaVu by 12.18$\times$ and 47$\times$, respectively.

cs.DC↗

arXiv2025

NetArena: Dynamic Benchmarks for AI Agents in Network Automation

As AI agents expand into high-stakes domains like network system operations, evaluating their real-world reliability becomes increasingly critical. However, existing benchmarks risk contamination due to static design, show high statistical variance from limited dataset size, and fail to reflect the complexity of production environments. We present NetArena, a dynamic benchmark generation framework for network applications. NetArena introduces a novel abstraction and unified interface that generalize across diverse tasks, enabling dynamic benchmarking despite the heterogeneity of network workloads. At runtime, users can generate unlimited queries on demand. NetArena integrates with network emulators to measure correctness, safety, and latency during execution. We demonstrate NetArena on three representative applications and find that (1) NetArena significantly improves statistical reliability across AI agents, reducing confidence-interval overlap from 85% to 0, (2) agents achieve only 13-38% average performance (as low as 3%) for large-scale, realistic queries, and (3) it exposes more fine-grained behaviors that static, correctness-only benchmarks miss. NetArena also enables use cases such as SFT and RL fine-tuning on network system tasks. Code is available at https://github.com/Froot-NetSys/NetArena.

cs.NI↗

arXiv2025

Approximation-First Timeseries Monitoring Query At Scale

Timeseries monitoring systems such as Prometheus play a crucial role in gaining observability of the underlying system components. These systems collect timeseries metrics from various system components and perform monitoring queries over periodic window-based aggregations (i.e., rule queries). However, despite wide adoption, the operational costs and query latency of rule queries remain high. In this paper, we identify major bottlenecks associated with repeated data scans and query computations concerning window overlaps in rule queries, and present PromSketch, an approximation-first query framework as intermediate caches for monitoring systems. It enables low operational costs and query latency, by combining approximate window-based query frameworks and sketch-based precomputation. PromSketch is implemented as a standalone module that can be integrated into Prometheus and VictoriaMetrics, covering 70% of Prometheus' aggregation over time queries. Our evaluation shows that PromSketch achieves up to a two orders of magnitude reduction in query latency over Prometheus and VictoriaMetrics, while lowering operational dollar costs of query processing by two orders of magnitude compared to Prometheus and by at least 4x compared to VictoriaMetrics with at most 5% average errors across statistics. The source code has been made available at https://github.com/Froot-NetSys/promsketch.

cs.DB↗

arXiv2025

Sketch Disaggregation Across Time and Space

Streaming analytics are essential in a large range of applications, including databases, networking, and machine learning. To optimize performance, practitioners are increasingly offloading such analytics to network nodes such as switches. However, resources such as fast SRAM memory available at switches are limited, not uniform, and may serve other functionalities as well (e.g., firewall). Moreover, resource availability can also change over time due to the dynamic demands of in-network applications. In this paper, we propose a new approach to disaggregating data structures over time and space, leveraging any residual resource available at network nodes. We focus on sketches, which are fundamental for summarizing data for streaming analytics while providing beneficial space-accuracy tradeoffs. Our idea is to break sketches into multiple `fragments' that are placed at different network nodes. The fragments cover different time periods and are of varying sizes, and are combined to form a network-wide view of the underlying traffic. We apply our solution to three popular sketches (namely, Count Sketch, Count-Min Sketch, and UnivMon) and demonstrate we can achieve approximately a 75% memory size reduction for the same error for many queries, or a near order-of-magnitude error reduction if memory is kept unchanged.

cs.NI↗

arXiv2025

Compact Data Structures for Network Telemetry

Collecting and analyzing of network traffic data (network telemetry) plays a critical role in managing modern networks. Network administrators analyze their traffic to troubleshoot performance and reliability problems, and to detect and block cyberattacks. However, conventional traffic-measurement techniques offer limited visibility into network conditions and rely on offline analysis. Fortunately, network devices -- such as switches and network interface cards -- are increasingly programmable at the packet level, enabling flexible analysis of the traffic in place, as the packets fly by. However, to operate at high speed, these devices have limited memory and computational resources, leading to trade-offs between accuracy and overhead. In response, an exciting research area emerged, bringing ideas from compact data structures and streaming algorithms to bear on important networking telemetry applications and the unique characteristics of high-speed network devices. In this paper, we review the research on compact data structures for network telemetry and discuss promising directions for future research.

cs.NI↗

arXiv2024

GEAR: An Efficient KV Cache Compression Recipe for Near-Lossless Generative Inference of LLM

Key-value (KV) caching has become the de-facto to accelerate generation speed for large language models (LLMs) inference. However, the growing cache demand with increasing sequence length has transformed LLM inference to be a memory bound problem, significantly constraining the system throughput. Existing methods rely on dropping unimportant tokens or quantizing all entries uniformly. Such methods, however, often incur high approximation errors to represent the compressed matrices. The autoregressive decoding process further compounds the error of each step, resulting in critical deviation in model generation and deterioration of performance. To tackle this challenge, we propose GEAR, an efficient KV cache compression framework that achieves near-lossless high-ratio compression. GEAR first applies quantization to majority of entries of similar magnitudes to ultra-low precision. It then employs a low rank matrix to approximate the quantization error, and a sparse matrix to remedy individual errors from outlier entries. By adeptly integrating three techniques, GEAR is able to fully exploit their synergistic potentials. Our experiments demonstrate that compared to alternatives, GEAR achieves near-lossless 4-bit KV cache compression with up to 2.38x throughput improvement, while reducing peak-memory size up to 2.29x. Our code is publicly available at https://github.com/HaoKang-Timmy/GEAR.

cs.LG↗

arXiv2024

Meili: Enabling SmartNIC as a Service in the Cloud

SmartNICs are touted as an attractive substrate for network application offloading, offering benefits in programmability, host resource saving, and energy efficiency. The current usage restricts offloading to local hosts and confines SmartNIC ownership to individual application teams, resulting in poor resource efficiency and scalability. This paper presents Meili, a novel system that realizes SmartNIC as a service to address these issues. Meili organizes heterogeneous SmartNIC resources as a pool and offers a unified one-NIC abstraction to application developers. This allows developers to focus solely on the application logic while dynamically optimizing their performance needs. Our evaluation on NVIDIA BlueField series and AMD Pensando SmartNICs demonstrates that Meili achieves scalable single-flow throughput with a maximum 8 μs latency overhead and enhances resource efficiency by 3.07$\times$ compared to standalone deployments and 1.44$\times$ compared to state-of-the-art microservice deployments.

cs.NI↗

arXiv2022

Enabling Efficient and General Subpopulation Analytics in Multidimensional Data Streams

Today's large-scale services (e.g., video streaming platforms, data centers, sensor grids) need diverse real-time summary statistics across multiple subpopulations of multidimensional datasets. However, state-of-the-art frameworks do not offer general and accurate analytics in real time at reasonable costs. The root cause is the combinatorial explosion of data subpopulations and the diversity of summary statistics we need to monitor simultaneously. We present Hydra, an efficient framework for multidimensional analytics that presents a novel combination of using a ``sketch of sketches'' to avoid the overhead of monitoring exponentially-many subpopulations and universal sketching to ensure accurate estimates for multiple statistics. We build Hydra as an Apache Spark plugin and address practical system challenges to minimize overheads at scale. Across multiple real-world and synthetic multidimensional datasets, we show that Hydra can achieve robust error bounds and is an order of magnitude more efficient in terms of operational cost and memory footprint than existing frameworks (e.g., Spark, Druid) while ensuring interactive estimation times.

cs.DB↗

arXiv2020

Sketchy With a Chance of Adoption: Can Sketch-Based Telemetry Be Ready for Prime Time?

Sketching algorithms or sketches have emerged as a promising alternative to the traditional packet sampling-based network telemetry solutions. At a high level, they are attractive because of their high resource efficiency and accuracy guarantees. While there have been significant recent advances in various aspects of sketching for networking tasks, many fundamental challenges remain unsolved that are likely stumbling blocks for adoption. Our contribution in this paper is in identifying and formulating these research challenges across the ecosystem encompassing network operators, platform vendors/developers, and algorithm designers. We hope that these serve as a necessary fillip for the community to enable the broader adoption of sketch-based telemetry.

cs.NI↗

arXiv2020

Unleashing In-network Computing on Scientific Workloads

Many recent efforts have shown that in-network computing can benefit various datacenter applications. In this paper, we explore a relatively less-explored domain which we argue can benefit from in-network computing: scientific workloads in high-performance computing. By analyzing canonical examples of HPC applications, we observe unique opportunities and challenges for exploiting in-network computing to accelerate scientific workloads. In particular, we find that the dynamic and demanding nature of scientific workloads is the major obstacle to the adoption of in-network approaches which are mostly open-loop and lack runtime feedback. In this paper, we present NSinC (Network-accelerated ScIeNtific Computing), an architecture for fully unleashing the potential benefits of in-network computing for scientific workloads by providing closed-loop runtime feedback to in-network acceleration services. We outline key challenges in realizing this vision and a preliminary design to enable acceleration for scientific applications.

cs.NI↗

arXiv2019

Privacy for Free: Communication-Efficient Learning with Differential Privacy Using Sketches

Communication and privacy are two critical concerns in distributed learning. Many existing works treat these concerns separately. In this work, we argue that a natural connection exists between methods for communication reduction and privacy preservation in the context of distributed machine learning. In particular, we prove that Count Sketch, a simple method for data stream summarization, has inherent differential privacy properties. Using these derived privacy guarantees, we propose a novel sketch-based framework (DiffSketch) for distributed learning, where we compress the transmitted messages via sketches to simultaneously achieve communication efficiency and provable privacy benefits. Our evaluation demonstrates that DiffSketch can provide strong differential privacy guarantees (e.g., $\varepsilon$= 1) and reduce communication by 20-50x with only marginal decreases in accuracy. Compared to baselines that treat privacy and communication separately, DiffSketch improves absolute test accuracy by 5%-50% while offering the same privacy guarantees and communication compression.

cs.LG↗

arXiv2019

Memory-Efficient Performance Monitoring on Programmable Switches with Lean Algorithms

Network performance problems are notoriously difficult to diagnose. Prior profiling systems collect performance statistics by keeping information about each network flow, but maintaining per-flow state is not scalable on resource-constrained NIC and switch hardware. Instead, we propose sketch-based performance monitoring using memory that is sublinear in the number of flows. Existing sketches estimate flow monitoring metrics based on flow sizes. In contrast, performance monitoring typically requires combining information across pairs of packets, such as matching a data packet with its acknowledgment to compute a round-trip time. We define a new class of \emph{lean} algorithms that use memory sublinear in both the size of input data and the number of flows. We then introduce lean algorithms for a set of important statistics, such as identifying flows with high latency, loss, out-of-order, or retransmitted packets. We implement prototypes of our lean algorithms on a commodity programmable switch using the P4 language. Our experiments show that lean algorithms detect $\sim$82\% of top 100 problematic flows among real-world packet traces using just 40KB memory.

cs.DS↗

arXiv2019

Enhancing the Privacy of Federated Learning with Sketching

In response to growing concerns about user privacy, federated learning has emerged as a promising tool to train statistical models over networks of devices while keeping data localized. Federated learning methods run training tasks directly on user devices and do not share the raw user data with third parties. However, current methods still share model updates, which may contain private information (e.g., one's weight and height), during the training process. Existing efforts that aim to improve the privacy of federated learning make compromises in one or more of the following key areas: performance (particularly communication cost), accuracy, or privacy. To better optimize these trade-offs, we propose that \textit{sketching algorithms} have a unique advantage in that they can provide both privacy and performance benefits while maintaining accuracy. We evaluate the feasibility of sketching-based federated learning with a prototype on three representative learning models. Our initial findings show that it is possible to provide strong privacy guarantees for federated learning without sacrificing performance or accuracy. Our work highlights that there exists a fundamental connection between privacy and communication in distributed settings, and suggests important open problems surrounding the theoretical understanding, methodology, and system design of practical, private federated learning.

cs.LG↗