Evolving big microservice architectures
•
1 recomendación•124 vistas
The document discusses evolving microservice architectures at Uber. It describes Uber's transition from a monolithic architecture to over 4,000 microservices with over 1,000 deploys per day. Key challenges discussed include testing changes, integrating services, handling multiple environments, and establishing common platforms and data models to support new use cases and services. The presentation emphasizes establishing layered, abstracted architectures with clear separations of concerns to facilitate engineering velocity at scale.
Recomendados
Recomendados
Más contenido relacionado
La actualidad más candente
La actualidad más candente (20)
Similar a Evolving big microservice architectures
Similar a Evolving big microservice architectures (20)
Más de Nikolay Stoitsev
Más de Nikolay Stoitsev (20)
Último
Último (20)
Evolving big microservice architectures
- 1. Evolving big microservice architectures Nikolay Stoitsev, Engineering Manager @ Uber Engineering
- 4. Dispatch API
- 5. A monolith? Code was consistently good Easy to reuse code Well tested Hard to deploy
- 6. Pull Request Pull Request Pull Request Pull Request Pull Request ? ? ? ? ?
- 7. Pull Request Pull Request Pull Request Pull Request Pull Request ? ? ? ? ? Unknown risk Unknown priority
- 8. Low Risk High Risk High Risk Med Risk High Risk Build
- 9. Low Risk High Risk High Risk Med Risk High Risk Build ?
- 10. Low Risk High Risk High Risk Med Risk High Risk Build
- 11. Low Risk / P0 High Risk / P1 High Risk / P0 Med Risk / P3 High Risk / P0 Build
- 12. Low Risk / P0 High Risk / P1 High Risk / P0 Med Risk / P3 High Risk / P0 Build CI / CD
- 13. Gradual rollout is important
- 14. Lesson learned: A monolith can scale a lot, but...
- 15. Dispatch API
- 21. 4000+ microservices 1K+ avg deploys a day
- 23. We want to introduce Uber Elevate
- 28. Dispatch Pricing Services Tax Calc Vehicles Trips Services Users Payments Services Receipts Task #1 OK Task #2 Task #3 Apache Kafka Apache Kafka Apache Kafka
- 29. Dispatch Pricing Services Tax Calc Vehicles Trips Services Users Payments Services Receipts Task #1 OK Task #2 Task #3 Task #4 Apache Kafka Apache Kafka Apache Kafka
- 30. Dispatch Pricing Services Tax Calc Vehicles Trips Services Users Payments Services Receipts Task #1 OK Task #2 Task #3 Task #4 Task #5Apache Kafka Apache Kafka Apache Kafka
- 31. Dispatch Pricing Services Tax Calc Vehicles Trips Services Users Payments Services Receipts Task #1 OK Task #2 Task #3 Task #4 Task #5 OKApache Kafka Apache Kafka Apache Kafka
- 32. Dispatch Pricing Services Tax Calc Vehicles Trips Services Users Payments Services Receipts Task #1 OK Task #2 Task #3 Task #4 Task #5 OK Task #6 Apache Kafka Apache Kafka Apache Kafka
- 33. How do we evolve big microservice architectures?
- 34. You can’t change what you can’t test
- 39. Multitenancy
- 40. Tenancy - set of data that is isolated from the other sets of data
- 42. Testing tenancy vs. Production tenancy
- 44. Audit tenancy
- 46. Dispatch Pricing Services Tax Calc Vehicles Trips Services Users Payments Services Receipts Task #1 OK Task #2 Task #3 Task #4 Task #5 OKApache Kafka Apache Kafka Apache Kafka
- 47. Platforms
- 48. Services are too concrete so it’s hard to support new use cases
- 49. Services need to support abstraction
- 50. AWS is a good example
- 55. Hard shift in mentality in the way you design your services
- 56. Can you sell it outside of the company?
- 57. Tax Calc Payments Pricing
- 58. Tax Calc Payments Pricing 3.14159
- 59. Tax Calc Payments Pricing 3.14159 “3.14”
- 60. Tax Calc Payments Pricing 3.14159 “3.14” 314159
- 64. Data extensions = schema + payload
- 65. How to do it? Schema service Repo Library ...
- 67. … but there is still business specific logic.
- 68. Dispatch Pricing Services Tax Calc Vehicles Orders Platform Users Payments Services Receipts Where to put it? Business logic Apache Kafka Apache Kafka Apache Kafka
- 69. No common understanding of what should go where
- 70. How do we call code with business logic, presentation logic, and SQL queries all over the place?
- 71. We fix spaghetti code with separation of concerns
- 74. Users Vehicles Order Platform Pricing Tax Calc Payments Receipts Rides Elevate Web Services Mobile Backend
- 75. Why?
- 76. Common language
- 77. Clear separation of concerns
- 79. API and functionality discoverability
- 81. Users Vehicles Order Platform Pricing Tax Calc Payments Receipts Rides Elevate Web Services Mobile Backend Generic Specific
- 82. Users Vehicles Order Platform Pricing Tax Calc Payments Receipts Rides Elevate Web Services Mobile Backend Breaks everything Breaks specific thing
- 83. Users Vehicles Order Platform Pricing Tax Calc Payments Receipts Rides Elevate Web Services Mobile BackendPresentation LoB Biz Infra Core
- 84. https://philcalcado.com/2018/09/24/services_layers.html Users Vehicles Order Platform Pricing Tax Calc Payments Receipts Rides Elevate Web Services Mobile Backend Clay Rocks
- 85. Users Vehicles Order Platform Pricing Tax Calc Payments Receipts Rides Elevate Web Services Mobile BackendPresentation
- 86. Users Vehicles Order Platform Pricing Tax Calc Payments Receipts Rides Elevate Web Services Mobile Backend LoB
- 87. Users Vehicles Order Platform Pricing Tax Calc Payments Receipts Rides Elevate Web Services Mobile Backend Biz Infra
- 88. Users Vehicles Order Platform Pricing Tax Calc Payments Receipts Rides Elevate Web Services Mobile Backend Core
- 89. Clear API
- 91. End result: Layered multitenant architecture built with abstract platforms and a common data model
- 92. Users Vehicles Order Platform Pricing Tax Calc Payments Receipts Rides Web Services Mobile Backend
- 93. Users Vehicles Order Platform Pricing Tax Calc Payments Receipts Rides Elevate Web Services Mobile Backend
- 94. Q&A Feedback:
- 95. Proprietary © 2019 Uber Technologies, Inc. All rights reserved. No part of this document may be reproduced or utilized in any form or by any means, electronic or mechanical, including photocopying, recording, or by any information storage or retrieval systems, without permission in writing from Uber. This document is intended only for the use of the individual or entity to whom it is addressed and contains information that is privileged, confidential or otherwise exempt from disclosure under applicable law. All recipients of this document are notified that the information contained herein includes proprietary and confidential information of Uber, and recipient may not make use of, disseminate, or in any way disclose this document or any of the enclosed information to any person other than employees of addressee to the extent necessary for consultations with authorized personnel of Uber.