Agile sre fly beyond the clouds olx core went aws- Wojciech Krysmann
0 likes•40 views
The document outlines OLX's migration to AWS through Agile Site Reliability Engineering (SRE) methodologies, detailing a strategic approach to infrastructure evolution. It emphasizes key lessons learned, including the importance of rituals, communication, and structured pull requests for effective collaboration. The migration process involves a logical slicing of systems, starting with less critical components before addressing the monolith, with a focus on measuring outcomes and making adjustments as necessary.
Agile sre fly beyond the clouds olx core went aws- Wojciech Krysmann
- 1. Agile SRE fly beyond the Clouds OLX core went AWS Wojciech Krysmann November 2018
- 2. /wkrysmann
- 4. 4 Horizontals New Ventures CarsReal Estate Verticals
- 7. Agile
- 8. “PEOPLE DON’T KNOW WHAT THEY WANT UNTIL YOU SHOW IT TO THEM” STEVE JOBS
- 10. Rituals
- 11. epic As a System Engineer, I want to run OLX.xx on AWS story As a System Engineer, I want to build cache cluster for olx.zz task Deploy memcache in elasticache bug App servers can’t connect to cache cluster Optional Board
- 12. story-7 Description In order to make a presentation easier to understand slides needs to be prepared. Proposed solution ● Use google slides, as you don’t have to back it up ● Use official template from shared resources ● Find new images in your photos resources Acceptance criteria What needs to be done: Presentation needs to consist of at least 40 slides. Use as much images, and as least text as possible What is out of scope: Presentation plan is out of scope. It will be done in story-2 Table of content will be created separately As a DevOpsDays speaker, I want to have a slide deck to visualise my thoughts
- 13. SRE
- 14. Solve infrastructure problems, like a developer would.
- 15. Share the pain, share the gain
- 16. embrace risk
- 21. Complexity?
- 22. Strategy ● Slice logically, decouple, and migrate one by one ● Than, tackle the monolith “cutover” at the end ● Start with least critical systems
- 24. 24 Timelines
- 26. Stories: 1. Design and build infrastructure 2. Prepare APP 3. Write images to two sources 4. A/B test new stack 5. Serve all new images from new infra 6. Background migrate old images 7. Close old infra
- 27. Design and build infrastructure
- 28. Prepare APP
- 29. Write images to two sources
- 30. A/B test new stack Serve all new images from new infra
- 31. Stories: 1. Design and build infrastructure 2. Prepare APP 3. Write images to two sources 4. A/B test new stack 5. Serve all new images from new infra 6. Background migrate old images 7. Close old infra 8. Revisit, adjust
- 32. OLXInfra as Code 45000 lines of code 500 tickets 1400 commits 6 people
- 33. Lessons learned
- 34. ● You can’t skip rituals ● Keep all communication on communicator ● Approver is not merging but author ● 1 PR = 1 Change = 1 user story ● if PR include Terraform code, than there is a plan as a PR comment ● if PR include Ansible code, there is a --check --diff output included ● If PR include Packer code, build it on dev first
- 35. ● Plan and balance between available time and feature set ● Take shortcuts sometimes ● Focus on goals ● Identify what can work bad in new environment