SlideShare une entreprise Scribd logo
1  sur  22
Télécharger pour lire hors ligne
A ML driven closed
loop automation using
ONAP
KATE
2020/08/31
Outline
• Introduction
• Recap
• ONAP design time and runtime.
• Basic concept of closed-loop automation.
• Telemetry and data analysis.
• How the ML improve the efficiency of system reactions.
• A use case- intelligent power management (ONS2019).
• Conclusions.
Introduction
• ONAP- 一個專為Telco市場量身打造的community 和 opensource project.
• A Linux foundation project.
• Led by AT&T.
• ONAP 的重點
• The concept.
• Design time and runtime disaggregate.
• Automate everything to reduce human errors.
• Driver- Telemetry and Data analysis.
• Use case- An Intelligent power management closed loop automation.
• Engagement.
Design-time and Run-time
• 對service deployment 而言,design-time and run-time 分離帶來
的好處:
• For developer:
• 把程式寫好就好。
• 包好。
• 再來model 執行環境(刻粿模)。
• For operator:
• 拿到粿模,丟進機器。
• 放上材料開始印粿。
• Telco industry中員工的分工是很絕對的。
Design-time
• Design-time:
• 描述服務:VNF, CNF, Web server, network, policy …etc.
• 建立License管理
• ACL
• SLA
• 日期之類的
• 你想的到的都有
• Model
• 透過正規化的語言描述服務- YAML (TOSCA) 。 **抱歉邱兄還是要當一下YAML engineer
• 到底要寫什麼呢?
• 設定的參數。
• Image 那邊來。
• 網路怎麼接。
• 理想是- Model 就像紅龜粿模,把服務一個一個印出來。
Runtime
• 顧名思意就是讓紅龜粿(扔到VIM上)跑起來~
• 但要全自動化製粿。
Service Model
(模)
Orchestrator
(機器)
Configuration
Hopper
(料)
VIM
(盤子~)
ONAP System Architecture
Design time Rune-time
Concept of Closed-loop Automation
• 就像車神~壓到石頭會自己把車騎正
• 或扶正!咩撲QQ
• 讓系統可以感知到 Service 的狀態。
• 讓系統可以藉由系統的回潰自動修改Service 讓系統恢復正常。
Telemetry
Automation
system
ServicesFeedbacks
Alerts/ Events
Configurations
Logs and
improvements
Administrators
ONAP Closed-loop Automation Example (in reality)
VIM
App1 App2
DCAE Policy
SO
Multicloud
SDNC
APPC
Telemetry
Event
Actions
Configs
Real NFV commands
Data Collection, Analytic, and Events (DCAE)
• Data Collection
• 不管從那裡來都可以。
• 一律送到MQ來處理。
• Protocol 主要有
• SNMP.
• VES. ß 這有點難搞…煩。
• 你自己寫也可以。
• Analytic module.
• 並沒有定義一定要用誰,過去使用過CDAP, PNDA…etc.
• 我是用python faust 啦。
• Events.
• 發生事件了就把事件往MQ丟就是了。
DCAE Architecture
DCAE platform
Consul
DCAE runtime
Cloudify
Policy
handler
Deployment
handler
Config
binding
service
PolicyCLAMP
DMaaP
SNMP/ VES
collector
HV-VES
TCA
(Threshold
analyzer)
xNFs
ML Brings the Benefits to Automation
• Traditional reactive prediction model.
• 頭痛醫頭,腳痛醫腳。
• 一直在對telemetry sampling 並評估下一段sampling period 的未來。
• 也許改改weight 可能很快。
• 但是也想想orchestration 所需要的時間。
• Proactive prediction and problem prevention.
• 預防頭痛,預防腳痛。
• 早早就把服務開好等著好戲上場(自動化)。
• 預先準備那些需要時間開啟的服務。
• 預先了解需求的成長決定投資的順序。
• 但我喜歡兩種和在一起使用。
ML Framework Integration to DCAE (Official)
Challenges
Training is an offline task. We don’t worry
about that.
Inference.
• 扯到很多東西。保證頭暈。
• Microservice management
• Route management
• Model management
Acumos
Inference micro services management
system.
SDC modeling 一切的東西。
ML Framework Integration to DCAE (土砲)
DCAE runtime
SNMP/ VES
collector
HV-VES
Inference
model
xNFs
DMaaP (Kafka/卡夫卡)
DB
conductor
DB
Offline
training
Model
publisher
Events
Policy
Events
Model
repo
想象中的admin UI
Telemetry data
Before Starting to Integrate.
• Behavior collection.
• 需要來點真正的資料,找到user behavior的pattern。
• 寫data training program,假裝自己是Data scientist.
• 對一對圖看看有沒錯誤。
• Measurement of system (xNF) performance under different
criteria.
• 建立baseline。
• 這樣才能正確計算出在每種configuration 系統應該要有的表現。
Building Model
這才是最難的- 取得真實的場域資料。
喔~我可以用incremental training。
喔~我可以用假資料
Balabala…
最後採用
LSTM + incremental training.
Expected Integration
Openstack
Bare metal servers
EPC EPC
GTP Input
GTP Output
Monitoring
ONAP
Training and logging
Actions and commands
What We did. (for monitoring and analytics)
DMaaP (Kafka)
VES
collector
xNFs and bare metal
Streaming
engine
DB
conductor
DB
Offline
training
Model
repo
ONAP
Policy
Trigger actions and
control VIMs and xNFs.
Events
Model
Telemetry
data stream
Data logging
Compare to Traditional Reactive Method
Reactive
最簡單的方法就是用moving average 來
track。
斜率的應用。
For example
• 報告學姐,2266日線和月線黃金交叉~
(scale out 當韭菜)。
• 發生死亡交叉就關幾個VNF囉。
慢,VNF可能準備好了事件都過去了。
Proactive (Prediction)
Prevent from something.
提早開服務增加availability。
快,但有時會做白工。
要付出overhead
Conclusions
• ONAP- Telco grade opensource project.
• A design concept for service operation and automation.
• 使用它的架構可以思考一下自己的operation flow 到底要長什麼
樣。
• 圓的還是扁的~anyway.
• 捲起袖子加入吧!
如何加入?
• Must—
• LFID
• ICLA
• CCLA
• 蒟蒻上面發ticket。
• Gerrit 上code。
• Openstack 的CICD process.
• Wiki 上面要看。
Thank You!
Q&A

Contenu connexe

Tendances

MongoDB World 2019: The Journey of Migration from Oracle to MongoDB at Rakuten
MongoDB World 2019: The Journey of Migration from Oracle to MongoDB at RakutenMongoDB World 2019: The Journey of Migration from Oracle to MongoDB at Rakuten
MongoDB World 2019: The Journey of Migration from Oracle to MongoDB at RakutenMongoDB
 
Spring Boot+Kafka: the New Enterprise Platform
Spring Boot+Kafka: the New Enterprise PlatformSpring Boot+Kafka: the New Enterprise Platform
Spring Boot+Kafka: the New Enterprise PlatformVMware Tanzu
 
Amazon kinesis와 elasticsearch service로 만드는 실시간 데이터 분석 플랫폼 :: 박철수 :: AWS Summi...
Amazon kinesis와 elasticsearch service로 만드는 실시간 데이터 분석 플랫폼 :: 박철수 :: AWS Summi...Amazon kinesis와 elasticsearch service로 만드는 실시간 데이터 분석 플랫폼 :: 박철수 :: AWS Summi...
Amazon kinesis와 elasticsearch service로 만드는 실시간 데이터 분석 플랫폼 :: 박철수 :: AWS Summi...Amazon Web Services Korea
 
나에게 맞는 AWS 데이터베이스 서비스 선택하기 :: 양승도 :: AWS Summit Seoul 2016
나에게 맞는 AWS 데이터베이스 서비스 선택하기 :: 양승도 :: AWS Summit Seoul 2016나에게 맞는 AWS 데이터베이스 서비스 선택하기 :: 양승도 :: AWS Summit Seoul 2016
나에게 맞는 AWS 데이터베이스 서비스 선택하기 :: 양승도 :: AWS Summit Seoul 2016Amazon Web Services Korea
 
Data Pipelines with Kafka Connect
Data Pipelines with Kafka ConnectData Pipelines with Kafka Connect
Data Pipelines with Kafka ConnectKaufman Ng
 
SDN입문 (Overlay and Underlay)
SDN입문 (Overlay and Underlay)SDN입문 (Overlay and Underlay)
SDN입문 (Overlay and Underlay)NAIM Networks, Inc.
 
VMware Advance Troubleshooting Workshop - Day 4
VMware Advance Troubleshooting Workshop - Day 4VMware Advance Troubleshooting Workshop - Day 4
VMware Advance Troubleshooting Workshop - Day 4Vepsun Technologies
 
Private cloud network architecture (2018)
Private cloud network architecture (2018)Private cloud network architecture (2018)
Private cloud network architecture (2018)Gasida Seo
 
[AWS Container Service] Getting Started with Cloud Map, App Mesh and Firecracker
[AWS Container Service] Getting Started with Cloud Map, App Mesh and Firecracker[AWS Container Service] Getting Started with Cloud Map, App Mesh and Firecracker
[AWS Container Service] Getting Started with Cloud Map, App Mesh and FirecrackerAmazon Web Services Korea
 
Introduction to Amazon Elastic File System (EFS)
Introduction to Amazon Elastic File System (EFS)Introduction to Amazon Elastic File System (EFS)
Introduction to Amazon Elastic File System (EFS)Amazon Web Services
 
Introduction to Amazon Web Services
Introduction to Amazon Web ServicesIntroduction to Amazon Web Services
Introduction to Amazon Web ServicesRobert Greiner
 
[G-Tech2015]次世代ファイアウォール -Cisco ASA with FirePOWER Services- によるセキュリティ対策[講演資料]
[G-Tech2015]次世代ファイアウォール -Cisco ASA with FirePOWER Services- によるセキュリティ対策[講演資料][G-Tech2015]次世代ファイアウォール -Cisco ASA with FirePOWER Services- によるセキュリティ対策[講演資料]
[G-Tech2015]次世代ファイアウォール -Cisco ASA with FirePOWER Services- によるセキュリティ対策[講演資料]Trainocate Japan, Ltd.
 
Microsoft Azure Security Overview - Microsoft - CSS Dallas Azure
Microsoft Azure Security Overview - Microsoft - CSS Dallas AzureMicrosoft Azure Security Overview - Microsoft - CSS Dallas Azure
Microsoft Azure Security Overview - Microsoft - CSS Dallas AzureAlert Logic
 
마이크로 서비스를 위한 AWS Cloud Map & App Mesh - Saeho Kim (AWS Solutions Architect)
마이크로 서비스를 위한 AWS Cloud Map & App Mesh - Saeho Kim (AWS Solutions Architect)마이크로 서비스를 위한 AWS Cloud Map & App Mesh - Saeho Kim (AWS Solutions Architect)
마이크로 서비스를 위한 AWS Cloud Map & App Mesh - Saeho Kim (AWS Solutions Architect)Amazon Web Services Korea
 
클라우드 상에서의 효율적인 데이터 보관 방법 - 김민형 클라우드 솔루션 아키텍트
클라우드 상에서의 효율적인 데이터 보관 방법 - 김민형 클라우드 솔루션 아키텍트클라우드 상에서의 효율적인 데이터 보관 방법 - 김민형 클라우드 솔루션 아키텍트
클라우드 상에서의 효율적인 데이터 보관 방법 - 김민형 클라우드 솔루션 아키텍트NAVER CLOUD PLATFORMㅣ네이버 클라우드 플랫폼
 
Ovs dpdk hwoffload way to full offload
Ovs dpdk hwoffload way to full offloadOvs dpdk hwoffload way to full offload
Ovs dpdk hwoffload way to full offloadKevin Traynor
 
Webinar: Accelerate Your Cloud Business With CloudHealth
Webinar: Accelerate Your Cloud Business With CloudHealthWebinar: Accelerate Your Cloud Business With CloudHealth
Webinar: Accelerate Your Cloud Business With CloudHealthCloudHealth by VMware
 

Tendances (20)

MongoDB World 2019: The Journey of Migration from Oracle to MongoDB at Rakuten
MongoDB World 2019: The Journey of Migration from Oracle to MongoDB at RakutenMongoDB World 2019: The Journey of Migration from Oracle to MongoDB at Rakuten
MongoDB World 2019: The Journey of Migration from Oracle to MongoDB at Rakuten
 
Spring Boot+Kafka: the New Enterprise Platform
Spring Boot+Kafka: the New Enterprise PlatformSpring Boot+Kafka: the New Enterprise Platform
Spring Boot+Kafka: the New Enterprise Platform
 
Amazon kinesis와 elasticsearch service로 만드는 실시간 데이터 분석 플랫폼 :: 박철수 :: AWS Summi...
Amazon kinesis와 elasticsearch service로 만드는 실시간 데이터 분석 플랫폼 :: 박철수 :: AWS Summi...Amazon kinesis와 elasticsearch service로 만드는 실시간 데이터 분석 플랫폼 :: 박철수 :: AWS Summi...
Amazon kinesis와 elasticsearch service로 만드는 실시간 데이터 분석 플랫폼 :: 박철수 :: AWS Summi...
 
2020.02.06 우리는 왜 glue를 버렸나?
2020.02.06 우리는 왜 glue를 버렸나?2020.02.06 우리는 왜 glue를 버렸나?
2020.02.06 우리는 왜 glue를 버렸나?
 
나에게 맞는 AWS 데이터베이스 서비스 선택하기 :: 양승도 :: AWS Summit Seoul 2016
나에게 맞는 AWS 데이터베이스 서비스 선택하기 :: 양승도 :: AWS Summit Seoul 2016나에게 맞는 AWS 데이터베이스 서비스 선택하기 :: 양승도 :: AWS Summit Seoul 2016
나에게 맞는 AWS 데이터베이스 서비스 선택하기 :: 양승도 :: AWS Summit Seoul 2016
 
Security on AWS
Security on AWSSecurity on AWS
Security on AWS
 
Data Pipelines with Kafka Connect
Data Pipelines with Kafka ConnectData Pipelines with Kafka Connect
Data Pipelines with Kafka Connect
 
SDN입문 (Overlay and Underlay)
SDN입문 (Overlay and Underlay)SDN입문 (Overlay and Underlay)
SDN입문 (Overlay and Underlay)
 
VMware Advance Troubleshooting Workshop - Day 4
VMware Advance Troubleshooting Workshop - Day 4VMware Advance Troubleshooting Workshop - Day 4
VMware Advance Troubleshooting Workshop - Day 4
 
Private cloud network architecture (2018)
Private cloud network architecture (2018)Private cloud network architecture (2018)
Private cloud network architecture (2018)
 
[AWS Container Service] Getting Started with Cloud Map, App Mesh and Firecracker
[AWS Container Service] Getting Started with Cloud Map, App Mesh and Firecracker[AWS Container Service] Getting Started with Cloud Map, App Mesh and Firecracker
[AWS Container Service] Getting Started with Cloud Map, App Mesh and Firecracker
 
[AWS Builders] Effective AWS Glue
[AWS Builders] Effective AWS Glue[AWS Builders] Effective AWS Glue
[AWS Builders] Effective AWS Glue
 
Introduction to Amazon Elastic File System (EFS)
Introduction to Amazon Elastic File System (EFS)Introduction to Amazon Elastic File System (EFS)
Introduction to Amazon Elastic File System (EFS)
 
Introduction to Amazon Web Services
Introduction to Amazon Web ServicesIntroduction to Amazon Web Services
Introduction to Amazon Web Services
 
[G-Tech2015]次世代ファイアウォール -Cisco ASA with FirePOWER Services- によるセキュリティ対策[講演資料]
[G-Tech2015]次世代ファイアウォール -Cisco ASA with FirePOWER Services- によるセキュリティ対策[講演資料][G-Tech2015]次世代ファイアウォール -Cisco ASA with FirePOWER Services- によるセキュリティ対策[講演資料]
[G-Tech2015]次世代ファイアウォール -Cisco ASA with FirePOWER Services- によるセキュリティ対策[講演資料]
 
Microsoft Azure Security Overview - Microsoft - CSS Dallas Azure
Microsoft Azure Security Overview - Microsoft - CSS Dallas AzureMicrosoft Azure Security Overview - Microsoft - CSS Dallas Azure
Microsoft Azure Security Overview - Microsoft - CSS Dallas Azure
 
마이크로 서비스를 위한 AWS Cloud Map & App Mesh - Saeho Kim (AWS Solutions Architect)
마이크로 서비스를 위한 AWS Cloud Map & App Mesh - Saeho Kim (AWS Solutions Architect)마이크로 서비스를 위한 AWS Cloud Map & App Mesh - Saeho Kim (AWS Solutions Architect)
마이크로 서비스를 위한 AWS Cloud Map & App Mesh - Saeho Kim (AWS Solutions Architect)
 
클라우드 상에서의 효율적인 데이터 보관 방법 - 김민형 클라우드 솔루션 아키텍트
클라우드 상에서의 효율적인 데이터 보관 방법 - 김민형 클라우드 솔루션 아키텍트클라우드 상에서의 효율적인 데이터 보관 방법 - 김민형 클라우드 솔루션 아키텍트
클라우드 상에서의 효율적인 데이터 보관 방법 - 김민형 클라우드 솔루션 아키텍트
 
Ovs dpdk hwoffload way to full offload
Ovs dpdk hwoffload way to full offloadOvs dpdk hwoffload way to full offload
Ovs dpdk hwoffload way to full offload
 
Webinar: Accelerate Your Cloud Business With CloudHealth
Webinar: Accelerate Your Cloud Business With CloudHealthWebinar: Accelerate Your Cloud Business With CloudHealth
Webinar: Accelerate Your Cloud Business With CloudHealth
 

Similaire à AI/ML-driven closed-loop automation using ONAP - Kate Hsuan

微博平台混合云实践 - Docker全架构
微博平台混合云实践 - Docker全架构微博平台混合云实践 - Docker全架构
微博平台混合云实践 - Docker全架构Chen Fei
 
服务器基准测试-叶金荣@CYOU-20121130
服务器基准测试-叶金荣@CYOU-20121130服务器基准测试-叶金荣@CYOU-20121130
服务器基准测试-叶金荣@CYOU-20121130Jinrong Ye
 
Ops as Code using Serverless
Ops as Code using Serverless Ops as Code using Serverless
Ops as Code using Serverless Rick Hwang
 
Data Analyse Black Horse - ClickHouse
Data Analyse Black Horse - ClickHouseData Analyse Black Horse - ClickHouse
Data Analyse Black Horse - ClickHouseJack Gao
 
从运维系统的开发谈安全架构设计
从运维系统的开发谈安全架构设计从运维系统的开发谈安全架构设计
从运维系统的开发谈安全架构设计mysqlops
 
MySQL压力测试经验
MySQL压力测试经验MySQL压力测试经验
MySQL压力测试经验Jinrong Ye
 
machine learning introduction
machine learning introduction machine learning introduction
machine learning introduction FEG
 
探索 ISTIO 新型 DATA PLANE 架構 AMBIENT MESH - GOLANG TAIWAN GATHERING #77 X CNTUG
探索 ISTIO 新型 DATA PLANE 架構 AMBIENT MESH - GOLANG TAIWAN GATHERING #77 X CNTUG探索 ISTIO 新型 DATA PLANE 架構 AMBIENT MESH - GOLANG TAIWAN GATHERING #77 X CNTUG
探索 ISTIO 新型 DATA PLANE 架構 AMBIENT MESH - GOLANG TAIWAN GATHERING #77 X CNTUGYingSiang Geng
 
openGauss DBMind.pdf
openGauss DBMind.pdfopenGauss DBMind.pdf
openGauss DBMind.pdfwot chin
 
在Nas的中心呼喊人工智慧 20171214
在Nas的中心呼喊人工智慧 20171214在Nas的中心呼喊人工智慧 20171214
在Nas的中心呼喊人工智慧 20171214Anderson Cheng
 
从林书豪到全明星 - 虎扑网技术架构如何化解流量高峰
从林书豪到全明星 - 虎扑网技术架构如何化解流量高峰从林书豪到全明星 - 虎扑网技术架构如何化解流量高峰
从林书豪到全明星 - 虎扑网技术架构如何化解流量高峰Scourgen Hong
 
05.wls调优
05.wls调优05.wls调优
05.wls调优Meng He
 
2014暑期訓練之Linux kernel power
2014暑期訓練之Linux kernel power2014暑期訓練之Linux kernel power
2014暑期訓練之Linux kernel power冠宇 陳
 
Times Ten Training
Times Ten TrainingTimes Ten Training
Times Ten TrainingLi Chen
 
云运维的理念和思维
云运维的理念和思维云运维的理念和思维
云运维的理念和思维昕 姚
 
Nodejs & NAE
Nodejs & NAENodejs & NAE
Nodejs & NAEq3boy
 
Sdn 之旅 open suse_asia_summit_20151206
Sdn 之旅 open suse_asia_summit_20151206Sdn 之旅 open suse_asia_summit_20151206
Sdn 之旅 open suse_asia_summit_20151206Fei Ji Siao
 

Similaire à AI/ML-driven closed-loop automation using ONAP - Kate Hsuan (20)

微博平台混合云实践 - Docker全架构
微博平台混合云实践 - Docker全架构微博平台混合云实践 - Docker全架构
微博平台混合云实践 - Docker全架构
 
服务器基准测试-叶金荣@CYOU-20121130
服务器基准测试-叶金荣@CYOU-20121130服务器基准测试-叶金荣@CYOU-20121130
服务器基准测试-叶金荣@CYOU-20121130
 
Ops as Code using Serverless
Ops as Code using Serverless Ops as Code using Serverless
Ops as Code using Serverless
 
Data Analyse Black Horse - ClickHouse
Data Analyse Black Horse - ClickHouseData Analyse Black Horse - ClickHouse
Data Analyse Black Horse - ClickHouse
 
从运维系统的开发谈安全架构设计
从运维系统的开发谈安全架构设计从运维系统的开发谈安全架构设计
从运维系统的开发谈安全架构设计
 
MySQL压力测试经验
MySQL压力测试经验MySQL压力测试经验
MySQL压力测试经验
 
machine learning introduction
machine learning introduction machine learning introduction
machine learning introduction
 
AI/ML O-RAN Cloud-Native Automation
AI/ML O-RAN Cloud-Native AutomationAI/ML O-RAN Cloud-Native Automation
AI/ML O-RAN Cloud-Native Automation
 
Storm
StormStorm
Storm
 
LabView with Lego NXT
LabView  with Lego NXTLabView  with Lego NXT
LabView with Lego NXT
 
探索 ISTIO 新型 DATA PLANE 架構 AMBIENT MESH - GOLANG TAIWAN GATHERING #77 X CNTUG
探索 ISTIO 新型 DATA PLANE 架構 AMBIENT MESH - GOLANG TAIWAN GATHERING #77 X CNTUG探索 ISTIO 新型 DATA PLANE 架構 AMBIENT MESH - GOLANG TAIWAN GATHERING #77 X CNTUG
探索 ISTIO 新型 DATA PLANE 架構 AMBIENT MESH - GOLANG TAIWAN GATHERING #77 X CNTUG
 
openGauss DBMind.pdf
openGauss DBMind.pdfopenGauss DBMind.pdf
openGauss DBMind.pdf
 
在Nas的中心呼喊人工智慧 20171214
在Nas的中心呼喊人工智慧 20171214在Nas的中心呼喊人工智慧 20171214
在Nas的中心呼喊人工智慧 20171214
 
从林书豪到全明星 - 虎扑网技术架构如何化解流量高峰
从林书豪到全明星 - 虎扑网技术架构如何化解流量高峰从林书豪到全明星 - 虎扑网技术架构如何化解流量高峰
从林书豪到全明星 - 虎扑网技术架构如何化解流量高峰
 
05.wls调优
05.wls调优05.wls调优
05.wls调优
 
2014暑期訓練之Linux kernel power
2014暑期訓練之Linux kernel power2014暑期訓練之Linux kernel power
2014暑期訓練之Linux kernel power
 
Times Ten Training
Times Ten TrainingTimes Ten Training
Times Ten Training
 
云运维的理念和思维
云运维的理念和思维云运维的理念和思维
云运维的理念和思维
 
Nodejs & NAE
Nodejs & NAENodejs & NAE
Nodejs & NAE
 
Sdn 之旅 open suse_asia_summit_20151206
Sdn 之旅 open suse_asia_summit_20151206Sdn 之旅 open suse_asia_summit_20151206
Sdn 之旅 open suse_asia_summit_20151206
 

AI/ML-driven closed-loop automation using ONAP - Kate Hsuan

  • 1. A ML driven closed loop automation using ONAP KATE 2020/08/31
  • 2. Outline • Introduction • Recap • ONAP design time and runtime. • Basic concept of closed-loop automation. • Telemetry and data analysis. • How the ML improve the efficiency of system reactions. • A use case- intelligent power management (ONS2019). • Conclusions.
  • 3. Introduction • ONAP- 一個專為Telco市場量身打造的community 和 opensource project. • A Linux foundation project. • Led by AT&T. • ONAP 的重點 • The concept. • Design time and runtime disaggregate. • Automate everything to reduce human errors. • Driver- Telemetry and Data analysis. • Use case- An Intelligent power management closed loop automation. • Engagement.
  • 4. Design-time and Run-time • 對service deployment 而言,design-time and run-time 分離帶來 的好處: • For developer: • 把程式寫好就好。 • 包好。 • 再來model 執行環境(刻粿模)。 • For operator: • 拿到粿模,丟進機器。 • 放上材料開始印粿。 • Telco industry中員工的分工是很絕對的。
  • 5. Design-time • Design-time: • 描述服務:VNF, CNF, Web server, network, policy …etc. • 建立License管理 • ACL • SLA • 日期之類的 • 你想的到的都有 • Model • 透過正規化的語言描述服務- YAML (TOSCA) 。 **抱歉邱兄還是要當一下YAML engineer • 到底要寫什麼呢? • 設定的參數。 • Image 那邊來。 • 網路怎麼接。 • 理想是- Model 就像紅龜粿模,把服務一個一個印出來。
  • 6. Runtime • 顧名思意就是讓紅龜粿(扔到VIM上)跑起來~ • 但要全自動化製粿。 Service Model (模) Orchestrator (機器) Configuration Hopper (料) VIM (盤子~)
  • 8. Concept of Closed-loop Automation • 就像車神~壓到石頭會自己把車騎正 • 或扶正!咩撲QQ • 讓系統可以感知到 Service 的狀態。 • 讓系統可以藉由系統的回潰自動修改Service 讓系統恢復正常。 Telemetry Automation system ServicesFeedbacks Alerts/ Events Configurations Logs and improvements Administrators
  • 9. ONAP Closed-loop Automation Example (in reality) VIM App1 App2 DCAE Policy SO Multicloud SDNC APPC Telemetry Event Actions Configs Real NFV commands
  • 10. Data Collection, Analytic, and Events (DCAE) • Data Collection • 不管從那裡來都可以。 • 一律送到MQ來處理。 • Protocol 主要有 • SNMP. • VES. ß 這有點難搞…煩。 • 你自己寫也可以。 • Analytic module. • 並沒有定義一定要用誰,過去使用過CDAP, PNDA…etc. • 我是用python faust 啦。 • Events. • 發生事件了就把事件往MQ丟就是了。
  • 11. DCAE Architecture DCAE platform Consul DCAE runtime Cloudify Policy handler Deployment handler Config binding service PolicyCLAMP DMaaP SNMP/ VES collector HV-VES TCA (Threshold analyzer) xNFs
  • 12. ML Brings the Benefits to Automation • Traditional reactive prediction model. • 頭痛醫頭,腳痛醫腳。 • 一直在對telemetry sampling 並評估下一段sampling period 的未來。 • 也許改改weight 可能很快。 • 但是也想想orchestration 所需要的時間。 • Proactive prediction and problem prevention. • 預防頭痛,預防腳痛。 • 早早就把服務開好等著好戲上場(自動化)。 • 預先準備那些需要時間開啟的服務。 • 預先了解需求的成長決定投資的順序。 • 但我喜歡兩種和在一起使用。
  • 13. ML Framework Integration to DCAE (Official) Challenges Training is an offline task. We don’t worry about that. Inference. • 扯到很多東西。保證頭暈。 • Microservice management • Route management • Model management Acumos Inference micro services management system. SDC modeling 一切的東西。
  • 14. ML Framework Integration to DCAE (土砲) DCAE runtime SNMP/ VES collector HV-VES Inference model xNFs DMaaP (Kafka/卡夫卡) DB conductor DB Offline training Model publisher Events Policy Events Model repo 想象中的admin UI Telemetry data
  • 15. Before Starting to Integrate. • Behavior collection. • 需要來點真正的資料,找到user behavior的pattern。 • 寫data training program,假裝自己是Data scientist. • 對一對圖看看有沒錯誤。 • Measurement of system (xNF) performance under different criteria. • 建立baseline。 • 這樣才能正確計算出在每種configuration 系統應該要有的表現。
  • 16. Building Model 這才是最難的- 取得真實的場域資料。 喔~我可以用incremental training。 喔~我可以用假資料 Balabala… 最後採用 LSTM + incremental training.
  • 17. Expected Integration Openstack Bare metal servers EPC EPC GTP Input GTP Output Monitoring ONAP Training and logging Actions and commands
  • 18. What We did. (for monitoring and analytics) DMaaP (Kafka) VES collector xNFs and bare metal Streaming engine DB conductor DB Offline training Model repo ONAP Policy Trigger actions and control VIMs and xNFs. Events Model Telemetry data stream Data logging
  • 19. Compare to Traditional Reactive Method Reactive 最簡單的方法就是用moving average 來 track。 斜率的應用。 For example • 報告學姐,2266日線和月線黃金交叉~ (scale out 當韭菜)。 • 發生死亡交叉就關幾個VNF囉。 慢,VNF可能準備好了事件都過去了。 Proactive (Prediction) Prevent from something. 提早開服務增加availability。 快,但有時會做白工。 要付出overhead
  • 20. Conclusions • ONAP- Telco grade opensource project. • A design concept for service operation and automation. • 使用它的架構可以思考一下自己的operation flow 到底要長什麼 樣。 • 圓的還是扁的~anyway. • 捲起袖子加入吧!
  • 21. 如何加入? • Must— • LFID • ICLA • CCLA • 蒟蒻上面發ticket。 • Gerrit 上code。 • Openstack 的CICD process. • Wiki 上面要看。