Accelerating DevOps with ChatOps
5 likes1,372 views
The document discusses the integration of ChatOps into DevOps practices to enhance team collaboration and streamline workflows. It highlights various use cases, including incident management and deployment, while also outlining limitations and advancements in ChatOps tools. The author emphasizes the importance of leveraging automation and continuous feedback to improve efficiency in IT teams.
Accelerating DevOps with ChatOps
- 1. Accelerating DevOps with ChatOps When, where and how to use ChatOps
- 2. Hello! I’m Vishwa Co-founder @ YellowAnt VishwaKK vishwa306
- 4. High performing IT teams experience 60 times less failure, recover from failure 168 times faster, deploy 30 times more frequently with 200 times shorter lead time - Rouan Wilsenach, ThoughtWorks
- 5. Did you know - Amazon deploys every 11.6 seconds!
- 7. Short Release cycles, rapid feedback, minimize MTTR
- 9. ChatOps Putting tools right in the middle of conversations - Jesse Newland, Github A collaboration model that connects people, tools, processes, conversations into a transparent workflow - Jason Hand, VictorOps
- 10. Ingredients
- 12. Bots
- 14. Situational awareness Connecting people and bringing people on the same page
- 15. Shopify Incident Management Bot Credits: Daniella N, Shopify
- 16. Leverage tribal knowledge The Team is Greater than the Sum of Its Members
- 17. Teams working in silos Credits: Jason Hand, VictorOps
- 18. Status Quo
- 19. APM incidents - Old methods Credits: Jason Hand, VictorOps
- 20. APM incidents - New methods Credits: Jason Hand, VictorOps
- 21. Use case - Deployment Credits: Rick Mak, Oursky
- 22. Use case - Incident Management Credits: Daniella N, Shopify
- 23. Use case - Incident Management Credits: Daniella N, Shopify
- 24. Use case - Checking Job status Credits: Rick Mak, Oursky
- 25. Use case - Build alerts Credits: Braintree
- 26. Use case - Infrastructure monitoring
- 27. Use case - Infrastructure monitoring
- 28. Did you know? Github recently fought off a massive DDoS attack with ChatOps!
- 29. ingress/egress anomaly(DDoS) notified to on-call team in Slack channel Bot command issued in Slack to withdraw BGP announcements over transit providers and announce AS36459 exclusively over Github’s links to Akamai Routes reconverged in the next few minutes and access control lists mitigated the attack at their border Continuous monitoring of transit bandwidth levels and load balancer response codes until full recovery Image Credits: S Kottler, Github
- 30. Use case - Environment provisioning Credits: Aaron Suggs, Kickstarter
- 31. Use case - Application management Credits: Stackstorm
- 32. Use case - Analytics Credits: Graphana
- 33. Use case - Task Management
- 34. ChatOps is not just for Devs!
- 35. Use case - Salesforce management
- 36. Use case - CRM
- 37. Use case - Marketing Credits: Dharmesh Shah, Hubspot
- 38. Use case - File sharing
- 39. How to implement ChatOps for your team?
- 41. Regex based command parsers Credits: Hubot
- 43. Hubot script - Example Add script to package.json Credits: Hubot
- 44. Hubot script - Example Credits: Hubot
- 45. NLP - Extracting intent and entities from commands
- 46. Limitations of Hubot, Lita, Err & Cog ● No set command syntax for every application → Syntax Overload ● Command with complex inputs are hard to implement and use ● Language! Application scripts need to be written in a single language(CoffeeScript, Ruby, Python etc.) ● Access Control - Difficult to implement access control ● Steep learning curve, mostly through repetition ● Lack of richer interaction elements like Buttons, Dialogs, Dropdowns etc. ● Lack of Cross-application interaction capability. For example, create a Github issue AND a JIRA Ticket in a single command
- 47. ChatOps 2.0 (2018-present) Leveraging rich UI elements to interface with bots
- 48. Meet YellowAnt A bot that lets you interface with your services and applications and helps you automate work from Slack, Microsoft Teams, Cisco Spark, FB Workplace
- 49. YellowAnt - application marketplace
- 50. Every application comprises of a set of commands
- 51. How YellowAnt commands get parsed maestro search_customer name “APPLE” country “US” Application Name Function Name Key Value parsing name: APPLE, country: US
- 52. Dialog inputs - better alternative to text commands
- 53. Dialogs - better alternative to text commands Application: Maestro Function: Create Issue Arguments: description: Webpage 500 error … issuetype: bug project: yellowant summary: Perform a performance test
- 54. Interaction - Command Flow Command in jira github maestro createissue issuecomment closeissue createrepo createissue create search_customer( name,country) get_collection create_customer Request Application API with input parameters Parse Construct and Return Message maestro search_customer name “APPLE” country “US” application: Maestro, function:search_customer Arguments: name: APPLE, Country: US
- 55. Architecture - Cloud OAuth User User-level Authentication App Server 1 App Server 2 App Server 3 App Container Python App Ruby App NodeJS App Python App Ruby App NodeJS App Some Apps from the Marketplace App Server - Standalone web application built on any platform App Container - Cluster of web applications hosted in one container YellowAnt Marketplace Application
- 56. Architecture - Hosted User User-level Authentication Node 3 Node 2 Python App Ruby App NodeJS App Java App Ruby App NodeJS App YellowAnt application Docker + integration application Dockers hosted in Kubernetes cluster Node 1 Load Balancer Persistent DB
- 57. Examples
- 58. Dialog inputs - better alternative to text commands
- 59. Command Object - transforming text and dialog commands { "data":{ "user": 4534, "function_name": "search_customer", "args": { "name":"APPLE", "country": "USA" }, "verification_token": "bgwreASFth09243rWE134804tnb", "application": 5639, "event": 5043867, "application_invoke_name": "maestro", "function": 74645, "event_type": "command" } }
- 60. How YellowAnt interacts with Applications search_customer( name,country) get_collection(customer) create_customer(name, address) Maestro API search_customer( name,country) get_collection(customer) create_customer(name, address) Maestro YellowAnt App Command in maestro search_customer name “APPLE” country “US” YellowAnt commands mapped to Maestro API Command Object
- 61. How to build applications on YellowAnt Step 1: Storyboarding - Identify use cases Step 2: Create YellowAnt Application in YellowAnt developer console Step 3: Define application functions and their arguments Step 4: Write code to implement user authentication, handle Command object and respond to user
- 62. Step 1: Storyboarding Sample YellowAnt application storyboard 1. Run Tasks with Maestro API 2. Fetch Analytics data 3. Query customer data
- 63. Step 2: Create YellowAnt application in developer console marqeta.yellowant.com/developers/
- 64. Step 3: Define application functions and their arguments
- 65. Step 4.1: User authentication Application 1. User visits application page 2. Application redirects to YellowAnt authentication page 3. User grants access 5. App exchanges auth code with long-lived token 4. Server returns authorization code 6. App creates user integration
- 66. Workflows Command workflow - Club multiple commands into one command Event workflow - Event in one application triggers actions in others
- 67. Command Workflows - Incident War room scenario
- 68. Event Workflows - demo video
- 69. Start small and simple, but dream big. Don’t be afraid to break things
- 70. Define goals and engagement metrics, get constant feedback
- 71. Be Consistent with your UX
- 72. Build confidence
- 73. Evangelize!
- 74. Keep it fun!
- 75. 2018 is the year of ChatOps!
- 76. Thank you!