Provides flexible computing capabilities for cloud services and large data processing, YC-backed start-up companies to launch points

The enterprise database market is huge, with the expertise of Oracle in this area, as well as the Cross-border giants of IBM (DB2) and Microsoft (SQL Server). They are similar to the open source database MySQL commonly used by small and medium-sized enterprises, belong to traditional relational database. It seems that the database market has developed very mature, the basic structure has been determined, so entrepreneurs have no chance?

In fact, the emergence and development of new technologies always bring new opportunities. As the development and maturation of object-oriented programming technology, DB4O is a pure object-oriented database. Cloud computing services and the rise of large data processing, so that traditional relational database in a number of application scenarios appear powerless. So with the support of YC, start-up company Citus data launched the analysis database Citusdb.

CITUSDB is a scalable analysis database, which is based on open source Object relational database PostgreSQL development. Users can access the CITUSDB master node as they would with a regular database, and then the master node assigns data and query requests to the work nodes of the database cluster, and the underlying schema is similar to Hadoop. In other words, CITUSDB is a database product that integrates SQL expressiveness, relational database performance, and the scalability and availability of Hadoop.

So, what's the advantage of it compared to the traditional database? To this end, we consulted the inauguration of the Zu Mingyu of the famous siege Lion. He was also the first to contact the new database, after a simple understanding of his own understanding:

Preliminary understanding is similar to hbase+pig such a set of systems, are known for large data real-time access work. The difference is that Citusdb is PostgreSQL two times based on object-relational databases, and Hbase+pig is based on Hadoop and HDFs. The big difference between them and traditional databases is that they all use a distributed architecture (like Hadoop) that manages the configuration node and separates the storage nodes.

The most direct disadvantage of this construction is that the management configuration node is a single point and can only be solved by the master/standby approach. The advantages are obvious, that is, the ability of distributed analysis and calculation, this is the traditional database can not be done. With distributed computing power, scalable or flexible computing is relatively easy, but also close to the so-called cloud computing, these are not the advantages of traditional databases.

At the same time, he added: "The traditional DB once query is similar to a thread completion, while the distributed storage mechanism will split the task into a similar m/r, each thread after the completion of the operation of the results to return to the user." ”

Of course, this is the @ Zu Mingyu after a short period of time to give a personal opinion, only for you in understanding Citusdb for reference. If you are really interested in this database, you can build your own environment test and then jump to conclusions.

Currently can be directly downloaded to the Citus data website to CITUSDB installation package, 8 nodes and the following can be used for free. For the time being, only the Ubuntu and Fedora Systems are supported and the Amazon Machine image version is about to be released.

(Responsible editor: Duqing first)