Distributed Basic Knowledge Collection

I. CAP

• Data Consistency (C), equivalent to all nodes accessing the same copy of the latest data;
• High availability of data updates (A);
• can tolerate network partitioning (P).

Concept: It is important to note that p:partition tolerance refers to the system works well despite physical network partitions. Many of the network said that P refers to the data partition, all his mother is nonsense, the actual should be if the physical network is isolated, can work properly.The simplest way to understand the cap theory is to imagine two nodes on each side of the partition. Allowing at least one node to update the state results in inconsistent data, i.e. the loss of C nature. If the node on the partition side is set to not be available for data consistency, the A property is lost. Unless two nodes can communicate with each other, both C and a are guaranteed, which can lead to the loss of P-Properties.

The characteristic of a distributed system is the ability to perform state switching under high latency or unreliable transmission conditions. If we want to ensure the reliability of the operation of the system, it is necessary to ensure the robustness of both the node and the network failure, but not all systems can meet the security capabilities we require. In this article, we will explore some of the design considerations for distributed databases and how they respond to the situation of network partitioning.

When a message is sent between two nodes, the IP network may arbitrarily delete, delay, reorder, or copy messages, so many distributed systems use TCP to prevent the reordering and copying of messages. But TCP/IP is still inherently asynchronous : The network will delay the message arbitrarily, and the connection may be cut off at any time. In addition, fault diagnosis is unreliable: it may not be possible to determine whether a node has failed, whether the network connection has been cut off, or whether the operation is slower than expected.

This failure of the message to be arbitrarily delayed or severed is called a network partition. Partitions can occur in a network of production environments for a variety of reasons: the pressure of garbage collection (GC), network card (NIC) failure, switch failure, configuration error, network congestion, and so on. Because of the partitioning, the cap theorem limits the maximum security capability that a distributed system can achieve. When a message is cut off, the "consistent" (CP) system rejects requests from certain nodes to maintain linear consistency. While the "available" (AP) system can handle requests on all nodes, linear consistency must be sacrificed, because different nodes may have different opinions about the order of operations. When the network is in good condition, the system can guarantee consistency and availability, but because there is always a partition in the real network, there is no system that is fully "consistent and available" (CA).

1.1 InfoQ, Cap theory 12 review: "Rules" changed: http://www.infoq.com/cn/articles/cap-twelve-years-later-how-the-rules-have-changed/1.2CP, CA system test, Ultimate version: Http://www.linuxeden.com/html/news/20131218/146570.html1.3 ACID, base:http://blog.sina.com.cn/s/blog_3fe961ae010139u6.html two. CDRT

2.1. Cdrt:conflict-free Replicated Data TYPES:HTTP://WWW.TUICOOL.COM/ARTICLES/A63EIU

2.2. CDRT Paper:http://hal.upmc.fr/file/index/docid/555588/filename/techreport.pdf 2.3 hbase in acid:http:// Archfan.sinaapp.com/2012/04/18/acid-in-hbase/6. Read and write separation--TODO

Distributed Basic Knowledge Collection