In-memory databases can be divided into relational memory databases and key-value memory databases. In actual applications, in-memory databases are mainly used with large relational databases such as Oracle or MySQL, focusing on performance.
It functions like a cache and does not pay attention to data integrity and data consistency.
Key-value-based in-memory databases are easier to use than relational ones, with better performance and scalability, so they are used more in applications than relational in-memory databases.
Compare the functional features of FastDB, Memcached and Redis mainstream in-memory databases.
FastDB features include the following:
1. FastDB does not support client-server architecture, so all applications using FastDB must run on the same host;
2. Fastdb assumes that the entire database exists in RAM, and optimizes the query algorithm and interface based on this assumption.
3. Fastdb has no database buffer management overhead and does not need to transfer data between the database file and the buffer pool.
4. The entire fastdb search algorithm and structure are based on the assumption that all data exists in memory, so the efficiency of data swapping will not be very high.
5. Fastdb supports transactions, online backup and automatic recovery after system crash.
6. Fastdb is an application-oriented database. The database table is constructed through the class information of the application.
FastDB cannot support the Java API interface, which makes it unsuitable to use FastDB in this application.
Memcached
Memcached is an open source cache server system based on Key-Value. It is mainly used as a data cache for the database and cannot be completely called a database.
The memcached API uses a thirty-two-bit cyclic redundancy check (CRC-32) to calculate the key value and then distributes the data on different machines. When the table is full, the next newly added data will be replaced using the LRU mechanism. Since memcached is usually only used as a caching system, applications using memcached need additional procedures to update the data in memcached when writing back to slower systems (such as back-end databases).
memcached has client development packages in multiple languages, including: Perl, PHP, JAVA, C, Python, Ruby, C#.
Redis
Redis is a high-performance key-value database. The emergence of redis has largely compensated for the shortcomings of keyvalue storage such as memcached, and can play a very good supplementary role to relational databases in some situations. It provides C, Java, Python, Ruby, Erlang, and PHP clients.
1.MySQL 5
As one of the most popular open source databases today, the MySQL database provides users with a relatively simple solution that is suitable for a wide range of application deployments and can reduce users' TCO. MySQL is a multi-threaded, Structured Query Language (SQL) database server. MySQL has high execution performance, fast running speed and is easy to use.
MySQL includes the following key advantages:
◆ Reliable performance and services MySQL provides early versions of all database server software to the public, which are tested for several months using open source before being released for production use.
◆ Easy to use and deploy MySQL's structural system is easy to customize and runs fast. Its unique multi-storage engine structure provides flexibility to enterprise customers, brings compactness and stability to the database management system, and is easy to deploy.
◆ Free access to source code allows you to access MySQL source code at any time, and its strategy ensures freedom and avoids locking into a certain company or platform.
◆ Cross-platform support MySQL is available on more than 20 different platforms, including major Linux systems, Mac OS X, Unix and Windows
◆ Reliable Development Power MySQL has a large user base and a highly qualified and experienced development team.
◆ Meets enterprise needs MySQL structure system is simple and easy to use, runs extremely fast, and can handle most application needs of enterprise databases.
On December 8, 2008, Sun Microsystems announced that it would officially provide MySQL 5.1 software to the outside world - this is an extremely important new version of MySQL, the world's most popular open source database. MySQL 5.1 GA version is now provided in the following three modes to meet the various special needs of different users:
◆MySQL Community Server - a free open source version of Sun's MySQL database. This GPL-licensed, full-featured software is targeted at individual technical users who do not require commercial support or preferential in-flight services.
Apache Three Musketeers: HBase, Cassandra, CouchDB. HBase has the most promising prospects because it has many developers and they are all top experts. Cassandra currently has a lot of negative voices. CouchDB is small but powerful, and has received many praises. CouchBase, which will be officially released, combines MemBase and CouchDB, which is very exciting.
HBase and Cassandra are both column-based databases modeled after Google’s BigTable, and they are both written in Java. Another similar database is HyperTable, which is used for some background analysis because it is written in C and is relatively fast. However, HyperTable is a bit fringe and not very popular. These column-based open source databases are currently one order of magnitude less than Google’s BigTable
CouchDB is a document database. Its biggest competitor is MongoDB. Both MongoDB and HBase adopt a master-slave server design. CouchDB's server distribution design is similar to Cassandra, Peer to Peer type. The master-slave server design is generally more consistent and belongs to the CP type in the CAP theory. CouchDB and Cassandra are generally considered to be event consistent and belong to the AP type in the CAP theory. But in fact, both MongoDB and Cassandra can be set to strong consistent or event consistent.
The databases mentioned above all support MapReduce. It seems that HyperTable supports non-primary key indexes. Both HBase and MongoDB with strong consistent configuration support the most basic locking (HBase single row locking, MongoDB single document locking), so transactions can be implemented, but the implementation is a bit complicated and inefficient. In terms of transactions alone, the current open source NoSQL database does not do a good job.
The biggest selling point of MongoDB is that it can perform many queries without building non-primary key indexes. However, MongoDB's server distribution design is really not flattering. It can be said to be the most ugly implementation among NoSQL databases.
There are many K-V databases, and the column-based databases and document databases mentioned above are actually K-V databases. The more popular purebred K-V databases include:
Memcached: Very popular, does not support persistence
VMWare's Redis: Very popular, used by Sina and Zhihu, CP type.
MemBase: Developed by many Memcached developers, using sqlite as the underlying storage. It is often used in social games, Zynga is using it, CP type.
Riak, the distributed implementation is more similar to CouchDB/Cassandra, AP type. Support MapReduce.
Linkin's Voldemort, a rare event consistent AP type in K-V.
TT, TC
Purely based on two-dimensional coordinate index is Neo4j. But now both MongoDB and CouchDB integrate this feature.
CouchOne, a company established by the developers of CouchDB, acquired MemBase, replaced its underlying sqlite with CouchDB and launched CouchBase, thus introducing MapReduce to support non-primary key indexes. CouchBase has not officially released the official version yet, but it will be soon. Although CouchDB is event consistent, the developers of CouchBase claim that CouchBase maintains the strong consistent feature of MemBase, and the specific implementation remains to be studied in the future.
If you look at it from a mature perspective, the more mature and very popular ones include CouchDB, Memcached, and Redis.
The above is the detailed content of What in-memory databases similar to Redis are available under Windows?. For more information, please follow other related articles on the PHP Chinese website!
![[Web front-end] Node.js quick start](https://img.php.cn/upload/course/000/000/067/662b5d34ba7c0227.png)
