Cloud Service Management: Why Machine Learning is Now Essential
3 likes•266 views
This document discusses how machine learning is essential for adaptive automation in cloud service management. It begins by providing context on how businesses are employing cloud, DevOps, and agile practices to drive digital transformation and agility. It then discusses how machine learning can provide insights to increase efficiency through automated noise reduction and task automation, avoid outages by detecting behavioral changes before users are impacted, and reduce mean time to resolution by identifying probable root causes and providing context in dynamic environments. The document emphasizes that machine learning applied to operational data can provide adaptive automation to support agility at scale.
Cloud Service Management: Why Machine Learning is Now Essential
- 1. © 2018 IBM Corporation Cloud Service Management and why Machine Learning is now essential June 21 2018
- 2. © 2018 IBM Corporation Agenda Part 1: - Cloud Service Management Part 2: - Machine Learning is essential (for adaptive automation) Part 3: - Wrap-up, Call to action & Q&A
- 3. Cloud enables digital transformation To transform, organizations are employing App Modernization, Hybrid, and DevOps Supporting agility at scale requires managing increasing data growth, complexity, and dynamic environments
- 4. Deliver, reliable, competitive applications Fast Business Reality & needs: • Agile Application Delivery • End user experience & reliability • Lean Operations Management Ops Goal: Fewer problem tickets, faster resolution Dev Goal: Faster time to market, reduce disruptions Dev Test Stage Prod Sto p Shift Right Shift Left virtually every application & service will incorporate AI, Gartner10yrs of the top 20 companies in every industry will be disrupted in the next 3 years1/3 of apps must be refactored to move to cloud99% LoB Executive Application Owner Application Developer Chief Information Officer IT Operations Manager IT Operations Engineer
- 5. Business Imperatives are Driving Faster Change Agility depends on DevOps practices and Cloud-Enabled Process Innovation Systems of Record Operational Excellence Systems of Engagement Transformation & Differentiation Agile Management Traditional Management Traditional Model Agile Model Some, big IT projects Many, small 2-3 years Time to go live 2-3 months Lower Change rate Higher Centralized Governance Decentralized Cloud-ready, on-prem Tools Cloud-Native ITIL, CMMI Processes DevOps, Lean Hybrid Ops Hybrid Apps Source: The agile CIO: Mastering digital disruption. http://blog.kpmg.ch/the-agile-cio-mastering-digital-disruption/ 5
- 6. Process, Tools and Culture Growing an Agile organization requires adaptation across the organization . Process Tools & Technology Culture • Adjust processes to enable Agility • Continued High Availability and Performance • Built-to-Manage Approach • Integrate Cloud Service Management toolchain with existing ITSM capabilities • Implement New Tools (ChatOps, Runbook Automation, etc.) • Orient on Application Agility and shared success (DevOps) • Transition to New Roles (i.e. Site Reliability Engineer, First Responder) • Transition to Proactive monitoring (Analytics) 6
- 7. Enterprise DevOps Adoption “The Future is already here, it is just unevenly distributed” – William Gibson 7 New DevOps Startup • Full Stack Engineers • Highly Collaborative • Informal and Agile • Focused and Independent Enterprise Business Reality: • Some Agile Applications • Some Legacy Applications • Adopting Cloud Operating Model • Mix of Traditional and Cloud IT Service Management (ITIL) • Process Oriented • Resistant to Change Cloud Service Management • Service Oriented • Dynamic and Agile L1 Ops L2 Ops SME Site Reliability Engineer First Responder DevOps/SME
- 8. Hybrid Cloud Management enables the transformation journey Theme Value Digital transformationAgility Adaptive Automation Select and manage the right cloud path for you Manageable, secure DevOps delivered at scale Recognize and respond to dynamic environments Flexibility
- 9. Cognitive Data Scientist learns, decides, improves AI to speed problem determination
- 10. © 2018 IBM Corporation Agenda Part 1: - Cloud Service Management Part 2: - Machine Learning is essential (for adaptive automation) Part 3: - Wrap-up, Call to action & Q&A
- 11. Adaptive automation of IT professionals agree: we will be overwhelmed without automation. 70%
- 12. Proactive Predictive insights Adaptive automation Reactive Real-time analytics Adaptive Cognitively enhanced workflow Scale Complexity Recognize and respond to dynamic environments Adaptive Automation Recognize and respond to dynamic environments
- 13. Insights to increase efficiency • Automated noise reduction • Automation of complex tasks Insights to Avoid Outages • Automatically detect behavioural changes • Take action, before users are impacted Insights to reduce MTTR • Probable cause identification • Context, in dynamic environments 13 Adaptive Automation Machine learning, advanced analytics and cognitive technologies delivering automated value for Centralized IT Operations and DevOps teams
- 14. Insights to increase efficiencyInsights to Avoid Outages Insights from your Terabytes of Operational Data Machine Learning applied automatically to your performance data Automate, automate, automate with Machine Learning applied your event and performance data. Extend with Watson. Reactive Real-time analytics Proactive Predictive Insights Adaptive Cognitively Enhanced Workflows Insights to reduce MTTR "“Right there - visually - we saw proof that you can use machine learning to be able to identify root cause….. Everyone sat there in silence for three minutes.” David Nestic Technical operations manager, NBN Source: "After testing the cognitive monitoring solution (IBM Operations Analytics Predictive Insights) ..we saw a significant reduction in server incidents..Thanks to it we will have a platform that can help us act before an incident occurs” Jan Steen Olsen Executive Vice President and CTO, Danske Bank Source: “We live on the edge of control, trying to assure our systems and deal with ever-changing business and user requirements. To control costs, we need to keep operations lean by processing only actionable alarms”…….On average, we reduced 15% of the “noise” alarms.” Operations Leader, Fast Growing Canadian Telco
- 15. Correlated Event Groups Traditional Events Cisco ACI Docker Kubernetes OpenStack TADDM NOI VMware vCenter ITNM IBM ALM DNS REST Netcool Ops Insight - Event Clustering - Seasonal Analysis and Suppression - Weighted probable cause Machine Learning for Reactive Management Cisco ACI Cognitive Event MoM Collaboration & Automation ChatOps Notification Run Books
- 16. Correlated Event Groups Traditional Events Proactive Events Metrics Cisco ACI Docker Kubernetes OpenStack TADDM NOI VMware vCenter ITNM IBM ALM DNS REST Predictive Insights Netcool Ops Insight - AI driven Model selection - Variance Analysis - Dependency Determination - Dynamic Threshold - Event Clustering - Seasonal Analysis and Suppression - Weighted probable cause Machine Learning for Reactive and Proactive Management Cisco ACI Cognitive Performance MoM Cognitive Event MoM Collaboration & Automation ChatOps Notification Run Books
- 17. Advanced Analytics for Rapid Context 17 Agile Service Manager Dynamic Topology MoM
- 18. RESULT: Cognitive Manager of Managers across Event, Performance and Topology data Cognitive Data Scientist learns, decides, improves Sophisticated Seasonal Modelling Robust Statistical approaches (independent of data distribution) Multiple Anomaly Detection Algorithms Automatic Model Validation Long term learning (monthly/ yearly patterns) Mathematical Relationship Discovery Rapid analysis of highly dynamic environments Automated Runbooks User Domain knowledge Alert Mgmt & Collaboration Probable Cause Identification Context, in highly Dynamic Environments Automated Remediation Mean-Time-To-Identify (MTTI) Mean-Time-to-Know (MTTK) Automated Event Suppression & Incident Correlation Automated Early Detection Mean-Time-to-Fix and Verify Adaptive Automation Incident Management Example
- 19. © 2018 IBM Corporation Agenda Part 1: - Cloud Service Management Part 2: - Machine Learning is essential (for adaptive automation) Part 3: - Wrap-up, Call to action & Q&A
- 20. Patterns of behavior w/ Machine Learning Seasonality of environment behavior Abnormal behaviors that precursor events Predict to Get Ahead Augment the Process Cognitive Automated Ticket Creation and Routing Cognitive Process Automation with robotics and Watson guided advise Cognitive Process Automation for zero-touch automation with robotics and Watson embedded advise and next steps Simplify & Focus Pattern Analysis to Correlate & De-duplicate events Pattern Analysis for IT Operations Cognitive Network 360* Insights Real Time Federated Topology Augment Staff Cognitive Incident Advisor Cognitive Agent Assist Cognitive Knowledgebase w/ semantic search Cognitive Assistant for Change §Netcool Operations Insight §Agile Service Manager §Hadoop HDFS §Watson Data Platform (DSX) §Watson Explorer §Watson Discovery §Watson Knowledge Studio §IBM Operations Analytics – Predictive Insights §Netcool Operations Insight §RPA tools §Watson Explorer Semantic Analysis §Dynamic Automation §PASIR §Watson Discovery §Watson Knowledge Studio §Watson Assistant §Watson Conversation Services §Watson Explorer §Watson Discovery §Watson Knowledge Studio §Speech To Text / Text To Speech §Watson for Cyber Security §Qradar Watson Advisor CapabilityProducts/CloudServices Adaptive Automation Machine Learning and leveraging user experience
- 21. Predict to Get Ahead Augment the ProcessSimplify & Focus Augment Staff Adaptive Automation Call to action and Q&A Short Videos Predictive Capabilities § The Value Video § The Capability video IBM Marketplace: § Operations Analytics § Netcool Operations Insight § Application Performance Management Forrester Total Economic Studies § The Operations Management TEI § The Application Management TEI IT Operations Maturity Assessment § Questionnaire to get you thinking Find Out More
- 22. © 2018 IBM Corporation Thank you
- 23. Key Capabilities: Reduce MTTR With 2nd Gen Advanced Real-Time Event, Performance and Topology Analytics • Groups events that always occur together, providing increased context for faster resolution • Learns complex relationships across your applications and infrastructure and provides insights for potential root cause • Rapidly analyses multiple sources of topology to provide up-to- date service and topology views for context With 1st Gen capabilities for Rapid Problem Resolution • Big data search across all operational data, supplemented text derived insights and log monitoring Insights to reduce MTTR • Probable cause identification • Context, in dynamic environments Reactive Real-time analytics Insights from your Terabytes of Operational Data
- 24. Key Capabilities: Avoid Outages Utilise IBM’s advanced machine learning to proactive manage your critical application and infrastructure Solution automatically detects behavioural changes and provides insights to help root cause Operations can take corrective action, before critical services and users are impacted Solution has been successfully deployed and has dramatically reduced outages Insights to Avoid Outages • Automatically detect behavioural changes • Take action, before users are impacted Proactive Predictive insights Machine Learning applied automatically to your performance data 2nd Gen ITOA
- 25. Key Capabilities: Increase Efficiency Due to advances, Machine Learning can now automate many human decisions AND it scales and adapts – Reduces alert noise due to advances seasonal behaviour analysis – Reduces manual effort by utilising machine learning to set and maintain thresholds – Reduce tickets and manual effort, by automatically grouping events that always occur together Automatically analyses patterns in operational data to identify waste and automation opportunities Insights to Increase Efficiency • Automated noise reduction • Automation of complex tasks Increase Efficiency Automate, automate, automate with Machine Learning applied your event and performance data 1st and 2nd Gen ITOA