Analysis and comparison of characteristics of common NoSQL

3. Classification:

3.1 (most common key-value storage)

A. Temporary: In memory, can be quickly saved and read operations, the data may be lost

B. Persistent: Data is stored on disk, read-write is fast (no memory is fast), data is not lost

C. Both: Memory and hard disk simultaneously save data, can be fast read and write operation, the hard disk data is not lost, suitable for processing array type of data

3.2 (document-oriented (MONGODB,COUCHDB))

A. Storage mode: No table structure, similar to Jason.

B. Complex query conditions can be made

3.3 (Column-oriented database hbase)

A. Relational databases are row-oriented, read-write to a small amount of travel, and column-oriented databases, read and write to a large number of rows, and update all specific columns on-the-go.

B. High scalability (especially written)

C. Very difficult to apply

D. Advantages for large amounts of data updates and queries

3.1 Detailed: memcached (temporary key value storage):

A. High-speed response

B. Used as a cache (indexed by relational databases can also get high-speed response): When dealing with large amounts of data, he can act as a relational database intermediary, acting as a cache for high-speed response, and by quickly reading the data inside to compensate for the possibility of losing data.

C. Storage mode: Hash table storage

D. Communication: Simple text protocol, the operation of the data is similar to the value of saving and key values, can only manipulate the character type data (other formats of data can be processed by Jason to format, so that can be transferred across the platform).

E. Simple operation of data,

F.memcached Classic algorithm for adding servers: Consistent hashing:

G. Cons: Because of his data in memory, 1. So it is easy to cause data loss. WORKAROUND: Use it only to resolve the copy of the data, and put the original data in a safe place. 2.web service with a lot of (foreign sites will basically introduce memcached),

In both cases it's powerful: 1. Some Web pages need to execute some good SQL text, slow response, some Web page access rate is high, load large, (relational database method to solve this problem: Add temporary table, batch processing)

Tokyo Tyrant (persistent key value store, stored on hard disk)

1. Advantage: Although his data is stored on the silver plate, but its data on the operation of the hard disk IO processing independent, can achieve high-speed access (and do not have to worry about the loss of data)

2. It also has the advantage that it can choose the structure of the storage according to the type of data, so it has its own advantage in the query, not only can be like the relational database scope query can also carry on the key-value to carry on the query of the completely consistent condition.

3. Data type: A. Cache the database in key-value form to store B.

4. It takes replication as a standard function

5. Cons: Hard to install (Ruby)

Redis (Temporary/persistent key value store)

1. Fast processing of data in the form of arrays

2. Through the virtual memory can be exchanged with the operating system structure, through the memory of all the data stored in the key to achieve efficient processing.

3. Quote, GitHub, live with a smile. Statistics of the number of spectators

4,. Application: Web application in timeline form

5. Suitable for processing time series data

MongoDB (High performance and extensible)

1. Document-oriented

2. The structure of the relational database is well defined in advance, and it is quite impressive to have information stored without altering the table structure, but it would be more troublesome to deal with the data that would be stored if the table structure was to be changed in later processing.

3. Common relational database implementation (repeated table structure transformation), you can serialize the data through JSON, then save to the table, and then read the data when the anti-serialization, and then filter out the data they need can also be resolved, but this down time overhead will increase.

3. You do not need a database that defines a table structure, and you can also add indexes based on key values that can be used for relational database queries.

4. Form of data preservation: BSON (binary product of JSON) associates the value with the key.

5, he cannot make jion queries, but it can be nested to achieve the same effect.

6. Advantages: There is no table structure, you can flexibly specify the query criteria. For example, a regular expression query or a query for a specific array element can

7. Example application: Storage of survey data

Performance verification of several basic NoSQL databases:

A NoSQL relational database

1. Ways to handlersocket the MySQL database to NoSQL:

Characteristics:

1. High-speed completion of simple data insertion and read processing

2. You can use SQL and handlersocket to access the same data

3. The MySQL data can be accessed independently by protocol

4. Do not support the handling of things

Handlesocket Machine Composition:

It has direct access to the handler interface, but is typically accessed through the SQL layer (eliminating the need to parse SQL and lock table and unlock)

Insufficient: Only query data with indexes can be supported.

Summary: Insert data section: Faster than the general relational database, although not memcached fast. Read speed can be similar to Redis

Berkeley DB

1. Storage form: key-Multiple Value form

2. Network access is not supported, the program is accessed through the in-process API

3. Architecture and its simplicity,

4, between the in-memory database and the relational database

Choose four algorithms: B + number algorithm: Balanced two-fork tree, ordered by its key word storage, and hash algorithm

1. Fast access speed.

2. Save Hard disk space

3. Most systems use it to support multiple processes and things

4 An XML interface that can store XML data

5, compatible with some old UNIX databases (DBM)

Analysis and comparison of characteristics of common NoSQL