You Can’t Sacrifice Partition Tolerance

http://www.ruanyifeng.com/blog/2021/06/weekly-issue-161.html

CAP 定理解读（英文）

这篇文章从学术角度，详细解读分布式系统的 CAP 定理，写得很好。CAP 定理的核心含义就是，发生故障时，开发者必须选择，优先满足一致性还是可用性。

You Can’t Sacrifice Partition Tolerance

In which there are limits to the CAP conjecture.

07 Oct 2010

I’ve seen a number of distributed databases recently describe themselves as being “CA” –that is, providing both consistency and availability while not providing partition-tolerance. To me, this indicates that the developers of these systems do not understand the CAP theorem and its implications.

 

A Quick Refresher

In 2000, Dr. Eric Brewer gave a keynote at the Proceedings of the Annual ACM Symposium on Principles of Distributed Computing¹ in which he laid out his famous CAP Theorem: a shared-data system can have at most two of the three following properties: Consistency, Availability, and tolerance to network Partitions. In 2002, Gilbert and Lynch² converted “Brewer’s conjecture” into a more formal definition with an informal proof. As far as I can tell, it’s been misunderstood ever since.

So let’s be clear on the terms we’re using.

后面还有很多其他的内容

 

 

References (i.e., Things You Should Read)

1. Brewer. Towards robust distributed systems. Proceedings of the Annual ACM Symposium on Principles of Distributed Computing (2000) vol. 19 pp. 7—10 ↩

2. Gilbert and Lynch. Brewer’s conjecture and the feasibility of consistent, available, partition-tolerant web services. ACM SIGACT News (2002) vol. 33 (2) pp. 59 ↩ ↩² ↩³ ↩⁴

3. DeCandia et al. Dynamo: Amazon’s highly available key-value store. SOSP ‘07: Proceedings of twenty-first ACM SIGOPS symposium on Operating systems principles (2007) ↩

4. Fox and Brewer. Harvest, yield, and scalable tolerant systems. Hot Topics in Operating Systems, 1999. Proceedings of the Seventh Workshop on (1999) pp. 174—178 ↩

5. Brewer. Lessons from giant-scale services. Internet Computing, IEEE (2001) vol. 5 (4) pp. 46-55 ↩ ↩²