1.1 Development

Key point 1: The semantics are simple and clear

Tip 2: Concise ≠ code short

1) Key data structures, such as data sharding structures;

1.2 Testing

2.1 Docker stand-alone clusters

2.2 Local outsourcing self-service E2E

3.1 Write test examples

3.2 Code Access Instructions

4.1 For Submitter

4.1 For Reviewer

5.1 Backbone development

5.2 Trunk/Branch Release

6.1 Test scaffolding

6.2 Environmental Standardization

7.1 git-poison poisoning

7.2 poison release blocking

8.1 From development to go-live

8.2 Sub-module release

9.1 Efficiency tools and methods

Documentation: online UML diagrams/flowcharts/milestones for multiple people to edit together, Teambition project management Gantt charts, Aone’s requirements management and defect management, offline tools such as Xmind mind maps/draw.io flowcharts/OneNote.

T: Time-bound, with a clear deadline

Trade-off: What are the advantages and disadvantages? What challenges do you bring?

System Implementation: Components and Key Implementations, Core Ideas and Core Components, Where is the Soul?

Related Works: What else is there to do on this issue? Related systems compared? Different implementations, different emphases, different ideas?

9.2 Personal growth and pit treading

Code submission is as atomic as possible, i.e. indivisible features, fixes, or optimizations, and test code is submitted in the same patch as production code

Implement a scaffolding that can be run and then continuously add content

New people are often down-to-earth, forget to look up at the stars, just focus on hard work, do not think about the business value behind it, this hoe, that shovel, everywhere is a pit, is not flowering, time-consuming and laborious, low sense of achievement.

Good Case: “In the 8 jobs 32 depth iosize 4K scenario when verifying performance on the 100g network standard card CX6, the limit IOPS drops from 1.2 million to 1.1 million, which is an 8% difference in performance compared to the FIC card”

Day arch a pawn, merit is not donated, and encouragement is shared.

Reference Links:

[1] Close to unacceptable load boundary


[2] Software Engineering at Google


[3] Test left-shifted engineering practices in large distributed systems


[4] Best Practices for Code Review


[5] Design Docs at Google


[6] The wisdom of questioning


Alibaba Cloud Storage White Paper

With the rise of Alibaba Cloud, various technologies within the group began to use Alibaba Cloud as the only export, Alibaba Cloud became the technology base of Alibaba’s economy, and Alibaba Cloud’s “Pangu” storage also became the storage chassis of Alibaba economy. It is most appropriate to describe the connotation of this white paper as “stable, secure and high-performance, Pratt & Whitney intelligent new storage”. “Don’t be afraid of floating clouds to cover your eyes, and be at the highest level.” Pangea-based Alibaba Cloud Storage will continue to lead the global industry into the future era of “new storage”.

Click to read the original article for details.