Upgrade to Pro
— share decks privately, control downloads, hide ads and more …
Speaker Deck
Features
Speaker Deck
PRO
Sign in
Sign up for free
Search
Search
Elasticsearch 在智能运维领域的运用
Search
medcl
January 20, 2018
Technology
0
340
Elasticsearch 在智能运维领域的运用
分享Elasticsearch和X-Pack组件在智能运维领域的技术原理和应用实践,如非监督型机器学习在自动的异常检测、高级关联和分类、根源问题诊断、早期故障预测等方面的应用等。
medcl
January 20, 2018
Tweet
Share
More Decks by medcl
See All by medcl
Elastic Stack- Past, Present, & Future
medcl
0
54
A Spider Written in Golang
medcl
1
76
又一个爬虫
medcl
0
93
Introduction to Beats and extending Beats
medcl
0
100
Elasticsearch & Bigdata
medcl
2
230
Elastic Stack V5
medcl
0
100
Elastic Stack V5
medcl
0
75
基于Elastic Stack的数据探索与分析@QConBeijing2016
medcl
1
440
Introduction to Elasticsearch @ FOSSASIA2016
medcl
0
7.5k
Other Decks in Technology
See All in Technology
Kiro IDEのドキュメントを全部読んだので地味だけどちょっと嬉しい機能を紹介する
khmoryz
0
210
GitHub Copilot CLI を使いやすくしよう
tsubakimoto_s
0
110
GitHub Issue Templates + Coding Agentで簡単みんなでIaC/Easy IaC for Everyone with GitHub Issue Templates + Coding Agent
aeonpeople
1
260
Codex 5.3 と Opus 4.6 にコーポレートサイトを作らせてみた / Codex 5.3 vs Opus 4.6
ama_ch
0
220
登壇駆動学習のすすめ — CfPのネタの見つけ方と書くときに意識していること
bicstone
3
130
22nd ACRi Webinar - NTT Kawahara-san's slide
nao_sumikawa
0
120
SREのプラクティスを用いた3領域同時 マネジメントへの挑戦 〜SRE・情シス・セキュリティを統合した チーム運営術〜
coconala_engineer
2
780
OpenShiftでllm-dを動かそう!
jpishikawa
0
140
Cloud Runでコロプラが挑む 生成AI×ゲーム『神魔狩りのツクヨミ』の裏側
colopl
0
150
22nd ACRi Webinar - 1Finity Tamura-san's slide
nao_sumikawa
0
110
Exadata Fleet Update
oracle4engineer
PRO
0
1.1k
Greatest Disaster Hits in Web Performance
guaca
0
300
Featured
See All Featured
Groundhog Day: Seeking Process in Gaming for Health
codingconduct
0
99
AI Search: Implications for SEO and How to Move Forward - #ShenzhenSEOConference
aleyda
1
1.1k
Design and Strategy: How to Deal with People Who Don’t "Get" Design
morganepeng
133
19k
It's Worth the Effort
3n
188
29k
Navigating the moral maze — ethical principles for Al-driven product design
skipperchong
2
250
SEO in 2025: How to Prepare for the Future of Search
ipullrank
3
3.3k
More Than Pixels: Becoming A User Experience Designer
marktimemedia
3
330
The Cost Of JavaScript in 2023
addyosmani
55
9.5k
The Impact of AI in SEO - AI Overviews June 2024 Edition
aleyda
5
740
Leveraging LLMs for student feedback in introductory data science courses - posit::conf(2025)
minecr
0
160
Darren the Foodie - Storyboard
khoart
PRO
2
2.4k
We Have a Design System, Now What?
morganepeng
54
8k
Transcript
Elasticsearch 在智能运维领域的运用 Elastic 技术专家 曾勇
什么是智能运维?
None
人工智能!
落地!
我们具体聊聊运维的痛点!
服务器器、硬件、⽹网络 …
软件、服务、代码…
传感器器、设备、物联⽹网… Image Credit: https://www.flickr.com/photos/teco_kit/23908928999
每时每刻产⽣生⼤大量量的…
事件、⽇日志…
Metrics、指标…
期望得到的…
报表…
异常…
告警…
因为…
Improve Uptime Stability Visibility Reduce Errors Downtime Time to Resolution
你需要…
所有这些. Unstructured Machine Learning Query language Fast Highly available Secure
Enrichment Advanced Analytics Dashboards Scalable Alerting SaaS Log correlation APIs Visualizations Real-time Drill down Reports Data sources
Elastic 提供 所有这些. Unstructured Machine Learning Query language Fast Highly
available Secure Enrichment Advanced Analytics Dashboards Scalable Alerting SaaS Log correlation APIs Visualizations Real-time Drill down Reports Data sources
Elastic 为什什么不不⼀一样?
⽤用户⽆无处不不在
LOG MANAGEMENT MOBILE APM SYSTEM MONITORING TIME SERIES WEB MONITORING
ANOMALY DETECTION Elastic 不不在 <Gartner 魔⼒力力象限> 尽管
Search Analytics Numbers Text Logs Historical Metrics Real time Heuristic
Machine Learning 多样性 才是我们的⻓长处 不不过
回到话题
运维之监控! • 监控指标的收集 • 监控数据的存储 • 监控数据的分析 • 监控数据的告警
运维之监控! • 监控数据的存储 • 监控数据的分析 • 监控数据的告警
Metricbeat, Filebeat, Auditbeat & Logstash System • Linux • MacOS
• Windows • Perfmon Custom apps • JMX/Jolokia • PHP-FPM • Golang • Dropwizard Storage • Ceph Cloud • AWS • GCP • DigitalOcean Queues • Redis • Kafka • RabbitMQ Security • ArcSight Caches • Memcached Containers • Docker • Kubernetes Virtualization • vSphere Datastores • MySQL • PostgreSQL • MongoDB • Couchbase • Aerospike Network • Netflow • Packets Web servers • Apache • Nginx Other • HAProxy • Zookeeper • Prometheus • Graphite • Icinga … …
运维之监控! • 监控数据的分析 • 监控数据的告警
运维之监控! • 监控数据的分析
运维之监控! 人工智能 亦或 只能人工
也还行, 也就几千个指标! CPU Metrics
也还行, 也就几千个指标! 几万? CPU Metrics
None
Elastic Stack 可采集海量指标 • 爆炸!
ELASTIC 的人工智能,智能运维。
ELASTIC 的人工智能,智能运维。 ELASTIC 的机器学习。
先看看监控数据 • 三大类 – Logging – Tracing info – Metrics
都是时序型数据!
什么是时序型数据?
为什么使用时序型数据?
为什么使用时序型数据?
Bucketing
Bucket 的选择
监督型机器学习。
非监督型机器学习。
时序型指标 特征化,模型化!
让机器去帮你 监控海量指标,发现异常!
DEMO
总结 • 运维已进入精细化智能化时代 • AI 不会让运维失业 • 让机器做机器擅长的 • Elastic
让运维分析更简单
None
IT-OPS-KPI
IT-OPS-NETWORK
IT-OPS-SQL
关联分析