Shulou(Shulou.com)05/31 Report--

This article mainly explains "what are the characteristics of CAP theory". The content of the explanation is simple and clear, and it is easy to learn and understand. Please follow the editor's ideas to study and learn "what are the characteristics of CAP theory"?

As the basic theory of distributed system, CAP theory describes a distributed system in the following three characteristics:

Consistency (Consistency)

Availability (Availability)

Partition fault tolerance (Partition tolerance)

At most two of these features are satisfied. That's what the following picture depicts. Distributed systems either meet CA, or CP, or AP. Cannot satisfy CAP at the same time.

i. What is consistency, availability, and partition fault tolerance

Partition fault tolerance: when a node or network partition in a distributed system fails, the whole system can still provide services that meet consistency and availability. In other words, some failures do not affect the overall use.

In fact, when we design a distributed system, we will take into account the failures caused by bug, hardware, network and other reasons, so even if some nodes or networks fail, we require the whole system to continue to use.

(if you do not continue to use it, it is equivalent to only one partition, so there will be no subsequent consistency and availability.)

Availability: read and write operations can always be done normally. To put it simply, the client can always be accessed normally and get the normal response of the system. From the user's point of view, there will be no problems such as system operation failure or access timeout.

Consistency: any read operation after a distributed system completes a write operation should get the latest value written by the write operation. It is equivalent to requiring the nodes in the distributed system to maintain the consistency of data at all times.

ii. How to understand it?

If we guarantee the fault tolerance of the partition in advance, it also means that if a node fails, the user can continue to access it. In this case, the consistency and availability of users cannot be satisfied at the same time during the access process. Refer to the following figure:

As shown in the figure, it is assumed that the distributed system has two nodes, G1 and G2, and the initial values are both v0. Now there is a client that writes the value v1 to the system, assuming that node G1 is written directly. After writing, client reads this value, and then reads the G2 node.

Because the G2 node loses connection with the G1 node, the data on the G1 node has not been synchronized to the G2 node, so the client reads the value v0 before modification. This gives rise to a situation where consistency is not satisfied. It is equivalent to satisfying usability and losing consistency.

Similarly, if the system ensures strong consistency, then after client writes the G1 node, there is a problem in synchronizing the data from the G1 node to the G2 node. If the client reads the data from the G2 node again, the client will be in a waiting state all the time, because the nodes in the system

Data for synchronization, need to wait for synchronization before it can be used. This is tantamount to satisfying consistency and losing usability.

When considering multiple client access, consistency and availability can also be understood in this way: if client1 modifies a value to G1 and the write operation is not completed, client2 will read the value, reading G2 nodes, if consistency is to be satisfied.

Then you have to make client2 unavailable for the time being, and if you want client2 to use it, the data you get is not up-to-date and the system is not consistent.

III. CAP can not have both, how to choose:

(1) CA: priority is given to ensuring consistency and availability, and fault tolerance of partitions is abandoned. This also means giving up the expansibility of the system, and the system is no longer distributed, which goes against the original intention of the design.

(2) CP: priority is given to ensuring consistency and partition fault tolerance, while abandoning availability. It is a common practice in situations where high data consistency is required (such as zookeeper,Hbase). Once a network failure or message loss occurs, the user experience will be sacrificed and the user will gradually be able to access it after recovery.

(3) AP: priority is given to ensuring availability and partition fault tolerance, while giving up consistency. Cassandra in NoSQL is such an architecture. Like CP, giving up consistency does not mean that consistency is not guaranteed, but gradually becomes consistent.

Thank you for your reading, the above is the content of "what are the characteristics of CAP theory". After the study of this article, I believe you have a deeper understanding of the characteristics of CAP theory, and the specific use needs to be verified in practice. Here is, the editor will push for you more related knowledge points of the article, welcome to follow!

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.