Big Talk NoSQL

    read a book called << Big Data challenge >> Span style= "font-family: ' Times New Roman '" >nosql nosql nosql We're sure to think of something bigdata Big data, yes, Today's era is the age of big data. If the internet hadn't been so developed in previous years, perhaps no one would have heard the term. At the point of the story, I made a diagram to see the General Service-side architecture in the face of business development needs, the general trend of evolution:

so if the company's data volume grows to a certain scale, it can use NoSql. Well, finally, it leads to NOSQL This is the subject of today. NoSql can be understood as "Not only Sql", mainly refers to the non-relational, distributed, ACID -not-available database design pattern, emphasizing "key-value storage" similar to MAP, and "Document Storage". In the face of massive data,NoSql uses a weak type of data, using a simpler data model, which typically uses strings to represent all data types. But he may not have the same strong consistency as a relational database, and NoSQL is a final consistency. NoSql also avoids a lot of complex operations, such as the query of tables, basically simple assignment, value, etc.

Can achieve relatively difficult throughput. Now the database system can be described as colorful, to take a look at the picture:

In the introductionNoSQLbefore, to know the theoretical knowledge of some massive data,CAPtheory,C,(Consistency) Strong consistency,A (Available)availability,P (Partition tolerance)partition tolerance can be understood as the system can still accept requests in the presence of a network partition. Naturally, this gives us another theory.ACID,A (Atomic)Atomic Nature,C (Consistent)consistency,I (Isolation)isolation,D (durablitity)sustainability. There's one more important protocol.2PCtwo phase commit protocol, many distributed relational databases use this protocol to accomplish distributed transactions. Let's take a real look at the specificNOSQL.

K-v Database

First of all, let's see.Redis,redisis an open-source, advancedKey-value's database. of hisvaluenot only supportStringtype, and there areList,Setand other structures,Redismemory is used for data storage, and when data reaches a certain scale, it is persisted to a file or disk. Based onRedisfeatures, Sina Weibo is usingRedis, for applications such as Weibo, which have a clear structure and a large data size ,Redisof course it's perfect.

column-oriented -Column Database

When it comes to the column database, we have to mention his ancestor,Bigtable database, in the later diffraction of a lot of databases can see his shadow, the column database is focused on a column-oriented sparse storage, there is a concept of a column family, with the relational database of single-row, Not the same, compare the HBase database used by typical Hadoop .

Document Type Database

Document type data, I'm talking about 2 kindsof things here,MongoDB, and a CouchDB, and they have a lot in common. Storage is JSON Type data types, documents in the document database become a basic unit of data storage, therefore, the stored data can even be required to be unstructured, the document can be long, short, each with a similar to the"{""ID": $,"msg":"haha"}such aJSONThe format is saved in the database. Let's focus onMongoDB, inMongoDBThere is a similarity in theSQLthe operation of the query statement, but not with theSQLstatements are identical, such asDb,user,find (), MongoDBSupportMap/reducemodel,CouchDBspecificRestFul API, can be implemented withHttpThe request implements the operation. Both databases use theMap/reduce,technology improves the efficiency of data processing, which is useful for storing unstructured and semi-structured data.

Survival database

Survival database I have just heard of this noun, it is also difficult to understand, the official definition: The survival database uses the underlying data structure to store data that represents a graph, to gracefully render any type of data in a very convenient manner. Survival database now generally has 3 classes,neo4j database,graphdb Survival database,orientdb, Survival database query will mimic the traversal of the graph to find, through friends to find friends, and finally find a goal.

nosql 4 Span style= "Font-family:times New Roman" >2 l.z algorithm series, now it should have been many versions of it, is based on the interaction of the window, the core idea is to reuse the same string If the following string appears preceded by a repetition, a tag can be used to save a significant amount of compression.

Well, spent more than a word, the rationale, also played a review of the effect.

Big Talk NoSQL