This article describes Suning's AI‑driven end‑to‑end solution for massive time‑series monitoring, anomaly detection, forecasting with DeepAR, MQ‑RNN, MQ‑CNN, ensemble methods, root‑cause localization using Hotspot and Monte‑Carlo Tree Search, and the evolution of its large‑scale log analytics platform.
This article describes how Tubi’s data science team built a robust KPI forecasting system with Facebook’s Prophet, covering visualization dashboards, anomaly detection, feature engineering, PySpark deployment, and evaluation using Brier scores to improve business decision‑making.
This article explains how WeBank’s intelligent operations team combines LSTM‑based forecasting with Gaussian analysis, k‑means feature clustering, and probability‑density modeling to automatically detect and warn about anomalies in key business metrics, moving beyond traditional threshold‑based monitoring.
The article presents a complete workflow for predicting next‑day API request volumes by exploring per‑minute traffic data, handling missing values, applying seasonal decomposition, training an ARIMA model on the trend component, and generating confidence intervals to flag anomalous spikes.
This article examines the severe time‑critical pain points of bank database operations, explains why AI‑based intelligent ops are needed, describes the platform architecture, unsupervised algorithms (3σ, Isolation Forest, DBSCAN, Pearson, Apriori), and presents a real‑world case study that demonstrates anomaly detection, root‑cause analysis, and practical optimization recommendations.
In a detailed Q&A session, Minsheng Bank’s database specialist shares 16 practical insights on intelligent database operations, covering data storage choices, monitoring tools like Zabbix and Prometheus, anomaly detection, prediction modeling, SQL metric collection, log analysis, and the evolving role of DBAs in the era of AI‑driven ops.
This article details JD Logistics' journey to design and implement a massive, AI‑enhanced monitoring platform that handles over three million metrics across hundreds of warehouses, addressing challenges of scale, network complexity, frequent asset changes, and integrating AIOps for proactive fault detection and resolution.
This article details a deep‑learning‑driven, real‑time alert system that combines TensorFlow LSTM time‑series forecasting with Spark Streaming to achieve high‑coverage, low‑latency anomaly detection for large‑scale data‑ops environments, including data preprocessing, metric classification, model training, and deployment pipelines.
This article presents the 2019 GOPS Global Operations Conference talk by Gong Cheng, detailing how intelligent monitoring leverages multi‑dimensional anomaly detection, machine‑learning‑based alert merging, knowledge‑graph construction, and root‑cause analysis to automate and improve large‑scale IT operations.
The article describes Ctrip's Prophet platform, which combines Flink real‑time stream processing with TensorFlow deep‑learning models to provide intelligent, low‑latency anomaly detection, replacing traditional rule‑based alerts and addressing challenges such as holiday traffic and model scalability.
This article explains the evolution of Alibaba DAMO Academy’s time‑series anomaly detection technology, detailing its application to both machine and commercial data, the challenges of diverse data types, the new robust statistical models, automatic data classification, parameter recommendation, and real‑world case studies demonstrating improved accuracy and stability.
This article explains why many system alerts are poorly designed, describes the true purpose of alerts as actionable notifications, distinguishes business rule monitoring from reliability monitoring, and presents practical metrics, strategies, and simple anomaly‑detection algorithms to create high‑quality, actionable alerts for reliable operations.
This article summarizes the keynote on intelligent operations presented at the 13th GOPS Global Operations Conference, covering multi‑dimensional anomaly detection, smart alarm aggregation, the construction of an operations knowledge graph, and AI‑driven root‑cause analysis techniques for large‑scale server environments.
This article explains Elasticsearch X‑Pack's machine‑learning capabilities, covering supervised and unsupervised learning concepts, data preparation, task creation types, architecture components, data flow, result indices, and provides code examples for configuring and running ML jobs.
The article presents a graph‑based anomaly detection architecture that tackles black‑market resource switching by constructing complex user‑traffic networks, mining graph similarities, and applying multi‑dimensional strategies to achieve high‑accuracy detection while meeting timeliness, performance, and interpretability requirements.
This article details the design, architecture, and AI‑driven features of 58’s intelligent monitoring platform, explaining how multi‑dimensional data collection, predictive analytics, and smart alarm merging ensure continuous, automated observability across network, server, application, and business layers.
This article explains the fundamental flaws of typical alert systems, distinguishes between business rule and reliability monitoring, outlines essential metrics and strategies for effective alerts, and presents simple yet powerful anomaly‑detection algorithms to ensure alerts are actionable and reduce noise.
This comprehensive guide explores a wide range of anomaly detection techniques—including time‑series methods, statistical models, distance‑based approaches, tree‑based isolation forests, graph algorithms, behavior‑sequence Markov models, and supervised machine‑learning models—detailing their principles, formulas, and practical scenarios for detecting outliers in advertising, fraud, and system monitoring.
By applying XGBoost‑based regression models that are retrained daily on two‑week order data and tuned per sub‑package, the mobile‑game recharge monitoring system predicts 10‑minute order volumes, sharply cuts false alarms from hundreds to dozens, and delivers precise, scalable alerts for game operations.
This article introduces the concept of anomaly detection, outlines common application scenarios such as ELT pipelines, feature engineering, A/B testing, and fraud detection, and reviews various detection methods—including statistical models, machine learning, rule‑based logic, and density‑based techniques—while discussing practical implementation considerations.
This article explains why most alerts are poorly designed, defines actionable alerts, outlines monitoring objectives, discusses metric selection, and presents simple yet powerful algorithms for anomaly detection to improve system reliability and operational efficiency.
This article describes a monitoring system that leverages machine learning to perform threshold‑free, real‑time anomaly detection on macro business indicators such as network traffic and access volume, detailing its architecture, sample labeling, model training, and multi‑level alarm strategies.
The article examines how telemarketing, a data‑intensive industry, leverages various data‑mining methods—including supervised classification, clustering, operations research optimization, anomaly detection, and text mining—to improve lead selection, agent allocation, churn prediction, and voice analysis, while also outlining the key data‑talent roles needed for successful implementation.
This comprehensive guide explores anomaly (outlier) detection across diverse methods—including time‑series analysis, statistical tests, distance metrics, matrix factorization, graph approaches, behavior‑sequence modeling, and supervised machine‑learning models—highlighting their principles, formulas, and practical use cases such as fraud prevention and system monitoring.
This article describes how a machine‑learning‑driven monitoring system replaces fixed thresholds with personalized, anomaly‑based detection for business‑level metrics such as network traffic and access volume, detailing the architecture, sample labeling, model training, alarm grading, and operational benefits.
This article details Ctrip's journey of applying AI-driven AIOps to address application operation pain points, describing their evolution from manual scripts to intelligent automation, the implementation of anomaly detection, smart diagnosis, online/offline mixed deployment, and future considerations for scalable, cost‑effective operations.
This article describes how a large‑scale operations monitoring system at Ctrip evolved from rule‑based alerts to a deep‑learning‑driven anomaly detection pipeline using LSTM models and Flink streaming, achieving a ten‑fold reduction in alert volume while improving fault detection.
In this talk, Alibaba senior tech expert Wang Zhaogang explains how intelligent monitoring, powered by machine‑learning algorithms and multi‑metric analysis, addresses the challenges of diverse business scenarios, enhances anomaly detection, improves root‑cause analysis, and shapes the future of smart operations.
This article presents 360's AIOps project, detailing how AI-driven capacity forecasting, host classification, resource recycling, intelligent MySQL scheduling, anomaly detection, alarm convergence, and root‑cause analysis have saved millions, improved efficiency, and paved the way for a fully automated operations workflow.
International credit card fraud, a growing threat, can be more effectively identified by applying unsupervised autoencoder models, which outperform traditional rule‑based systems by tripling recall and increasing accuracy by 40%, while reducing maintenance costs and adapting to new fraud patterns.
360’s AI‑ops team shares a year‑long journey of turning massive operational data into intelligent solutions—covering background, their AIOps philosophy, practical modules like capacity forecasting, host classification, resource reclamation, smart MySQL scheduling, anomaly detection, alarm reduction, and root‑cause analysis—to dramatically improve cost, efficiency, and reliability.
TcpRT is a real‑time instrumentation and diagnostic system for Alibaba Cloud RDS that non‑intrusively collects TCP trace data, aggregates billions of records per day, applies statistical and Cauchy‑based anomaly detection, and pinpoints root causes across hosts, proxies, and network devices at massive scale.
A mathematician‑turned‑engineer recounts his 2015‑2022 path from undocumented recommendation systems at Tencent, through high‑precision security models, reinforcement‑learning game AI, quantum‑ML studies, to large‑scale AIOps time‑series anomaly detection, offering practical lessons for anyone transitioning into machine learning.
This article presents a comprehensive guide to constructing a data‑driven AIOps platform, detailing its architecture, core components such as time‑series forecasting, anomaly detection, and pattern clustering, and recommending open‑source projects and practical considerations for implementing intelligent operations in enterprises.
This article explains how 360’s network operations team uses time‑series analysis, statistical thresholds, EWMA, dynamic limits, and machine‑learning models such as K‑Means and Isolation Forest to automatically detect, locate, and remediate traffic anomalies across massive data‑center exits.
This article summarizes a two‑stage method for anomaly detection when only a few labeled anomalies and many unlabeled instances are available, detailing problem formulation, isolation‑forest‑based scoring, clustering of anomalies, weighted multiclass modeling, experimental validation, and real‑world URL attack applications.
This article chronicles the author’s transition from a mathematics background to machine learning, detailing early challenges, hands‑on projects in recommendation systems, security, and AIOps, and sharing practical insights on feature engineering, model evaluation, and large‑scale anomaly detection.
This article examines Baidu's practical applications of machine‑learning‑driven intelligent operations, detailing three real‑world scenarios, the challenges of KPI anomaly labeling, the design of an automated detection framework, evaluation results across multiple datasets, and broader insights for scaling AIOps in production environments.
The article recounts the author’s 2017‑onward journey building an intelligent operations platform at Tencent, detailing challenges such as legacy thresholds, AIOps talent shortage, and lack of frameworks, and explains how a two‑stage hybrid unsupervised‑supervised model was devised to automatically detect anomalies across millions of KPI time‑series, enabling scalable root‑cause analysis and cost optimization.
This article outlines the challenges of modern IT operations, explains why traditional methods fall short, and presents a detailed AIOps roadmap—including anomaly detection, localization, root‑cause analysis, and prediction—drawn from Professor Pei Dan’s research and real‑world examples.
This article explains how WeChat’s security team builds a scalable anomaly‑detection framework that partitions billions of user accounts, weights suspicious attributes, computes similarity graphs, and leverages Spark optimizations and graph‑partitioning techniques to efficiently identify malicious user clusters.
Tencent's SNG social platform team tackles billion‑scale traffic by integrating AI‑driven anomaly detection, multi‑dimensional monitoring, and decision‑tree based root‑cause analysis, turning complex backend architectures and massive alert volumes into streamlined, actionable insights for faster issue resolution.
The Metric Logic Tree automates business metric anomaly analysis by integrating heterogeneous data sources (Kylin, MySQL, Elasticsearch, Druid) with a three‑layer architecture—metric calculation, algorithmic analysis (waterfall and Gini‑coefficient methods), and a master‑worker computation service—that parallelizes queries, delivers immediate conclusions, and shortens decision cycles, as demonstrated in Meituan‑Dianping’s hotel‑travel operations.
This article describes how Ctrip's security team replaced rule‑based web attack detection with a Spark‑powered machine‑learning pipeline, detailing the system architecture, feature engineering using TF‑IDF, model training, evaluation, online deployment, and future enhancements to improve detection accuracy and performance.
In this comprehensive talk, Pei Dan outlines the technical and strategic roadmap for bringing AIOps to production, explains the challenges of anomaly detection, fault localization, root‑cause analysis and prediction, and demonstrates how to decompose complex operations problems into AI‑solvable tasks.
This article introduces the main types of data‑mining algorithms—classification, prediction, clustering, and association—explains supervised and unsupervised learning, and illustrates each with practical examples such as spam detection, tumor identification, wine quality assessment, fraud detection, recommendation systems, and authorship analysis.
This article describes how Netflix’s technical operations team built an automatic anomaly‑detection system based on the DBSCAN clustering algorithm to identify subtly failing servers from time‑series error‑rate data, evaluate its effectiveness, and discuss practical deployment considerations.
