Strategies to edit production data
1 like•409 views
The document discusses various strategies for safely editing production data, highlighting the challenges posed by the absence of internal tools and the potential for errors in raw SQL execution. Five key strategies are outlined: developing a manual review process, running scripts locally, executing scripts on existing servers, utilizing a task runner, and building a script runner service, each with its own advantages and drawbacks. The text emphasizes the importance of auditing, user-friendly interfaces, and effective configuration management for successful implementation.
Strategies to edit production data
- 1. Strategies to Edit Production Data Julie Qiu @jqiu25
- 2. Julie Qiu Senior Software Engineer, Spring @jqiu25
- 3. Why We Edit Production Data
- 4. Internal tools are not available.
- 5. Internal tools are not available. Edge cases exist.
- 6. Internal tools are not available. Edge cases exist. Fast and easy to make changes.
- 7. I want to run this query UPDATE products SET name=‘julies-product’ WHERE id=1
- 8. What I actually ran UPDATE products SET name=‘julies-product’
- 10. Let’s talk about strategies for safer editing.
- 11. Raw SQL Local scripts Existing server Task runner Script runner service Strategies for editing
- 12. How the strategy works What’s great What’s not great Things we’ll discuss
- 13. Strategy 1: Develop a Review Process for Manual Edits
- 14. I want to run this query UPDATE products SET name=‘julies-product’ WHERE id=1
- 15. 1. Add record to a spreadsheet Process for manual edits Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 16. Manual Queries Spreadsheet Name Julie Qiu Date 1/19/2018 Description Marketing team request Query UPDATE products SET name=‘juliies-product’ WHERE id=1 Reviewer Name Codanda Appachu Reviewer Status Approved / Changes Requested Reviewer Comments You spelled Julie wrong
- 17. 1. Add record to a spreadsheet 2. Reviewer approves the query Process for manual edits Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 19. 1. Add record to a spreadsheet 2. Reviewer approves the query 3. Run the query Process for manual edits Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 20. Run the query inside a transaction BEGIN; UPDATE products SET name=‘julies-product’ WHERE id=1; END; Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 21. • Easy to implement What’s great Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 22. • Easy to implement • Audit trail What’s great Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 23. • Easy to implement • Audit trail • Promotes the right behaviors What’s great Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 25. • Easy to make mistakes What’s not great Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 26. • Easy to make mistakes • Audit trail is at will What’s not great Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 27. • Easy to make mistakes • Audit trail is at will • Difficult to run long and complex logic What’s not great Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 28. Varying this query UPDATE products SET name=‘julies-product’ WHERE id=1
- 29. UPDATE products SET name=‘julies-product’ FROM brands WHERE brands.name=‘julies-store‘ AND products.status=‘active‘ AND products.name IS NULL A more complex version
- 30. Strategy 2: Run Scripts Locally
- 31. 1. Write a script Run scripts locally Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 32. Write the script with your ORM of choice UPDATE products SET name=‘product-1’ WHERE id=1 Script Runner Service Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 33. 1. Write a script 2. Connect to remote database Run scripts locally Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 34. 1. Write a script 2. Connect to remote database 3. Run the script Run scripts locally Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 35. python update_product_name.py --product_name=‘julies-product’ --product_id=1 --dry-run Run the script (but first do a dry-run!) Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 36. python update_product_name.py --product_name=‘julies-product’ --product_id=1 --dry-run Run the script (but do a dry-run!) Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 37. • Reusable What’s great Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 38. • Reusable • Easy to manipulate outputs What’s great Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 39. • Reusable • Easy to manipulate outputs • Access to common code What’s great Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 41. • Easy to make mistakes What’s not great Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 42. • Easy to make mistakes • Logs are only available locally What’s not great Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 43. • Easy to make mistakes • Logs are only available locally • Network disconnections What’s not great Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 44. Varying this query UPDATE products SET name=‘julies-product’ WHERE id=1
- 45. UPDATE products SET name=‘product-1’ WHERE id=1 UPDATE products SET name=‘product-2’ WHERE id=2 UPDATE products SET name=‘product-3’ WHERE id=3 Doing this 50 million times UPDATE …
- 47. Strategy 3: Run Scripts on an Existing Server
- 48. 1. Write a script Run on an existing server Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 49. 1. Write a script 2. Get the script onto a server Run on an existing server Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 50. 1. Write a script 2. Get the script onto a server 3. SSH and run inside a session Run on an existing server Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 55. • Ability to run long scripts What’s great Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 56. • Ability to run long scripts • Reliable network connectivity What’s great Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 57. • Ability to run long scripts • Reliable network connectivity • Infrastructure already exists What’s great Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 59. • Scripts can affect resources on your server What’s not great Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 60. • Scripts can affect resources on your server • Not user friendly What’s not great Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 61. • Scripts can affect resources on your server • Not user friendly • No persistent audit trail What’s not great Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 62. Strategy 4: Use a task runner
- 64. 1. Write a script Use a task runner Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 65. 1. Write a script 2. Code review and run tests Use a task runner Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 66. 1. Write a script 2. Code review and run tests 3. Input arguments and run Use a task runner Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 69. • Persistent audit logs What’s great Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 70. • Persistent audit logs • Code review and automated tests What’s great Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 71. • Persistent audit logs • Code review and automated tests • User interface What’s great Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 73. • Hard to manage credentials What’s not great Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 74. • Hard to manage credentials • Environments are not clearly separated What’s not great Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 75. • Hard to manage credentials • Environments are not clearly separated • Inputs are not verified What’s not great Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 77. Strategy 5: Build a Script Runner Service
- 79. 1. Write a script Script runner service Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 80. 1. Write a script 2. Code review and run tests Script runner service Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 81. 1. Write a script 2. Code review and run tests 3. Choose environment and run Script runner service Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 83. • Centralized configuration management What’s great Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 84. • Centralized configuration management • Separation of environments What’s great Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 85. • Centralized configuration management • Separation of environments • User friendly interface What’s great Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 86. • Parallelize and scale More things we can do Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 87. • Parallelize and scale • Preview results More things we can do Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 88. • Parallelize and scale • Preview results • … up to you to customize! More things we can do Local Scripts Existing Server Task Runner Script Runner Service Local Scripts Raw SQL Raw SQL Raw SQL
- 89. Which strategy should I use?
- 90. Centralize Configs Separate Environ User Interface Audit Trail Code Review Long Scripts Complex Logic Data Editing Testing Benefits … Raw SQL Existing Server Local Scripts Script Runner Service Task RunnerTask Runner
- 91. Make something useable* *because engineers are also people!
- 92. Invest the effort – it’s worth the cost.