YunTable Development Diary (8) - to talk about the role of distributed database (reprint)

Source Address: http://peopleyun.com/?p=756

 

After the release YunTable0.1 version, I will the good news and I have a job in China Mobile shares Tongji students, he first of all to congratulate me, but he did not understand distributed databases such as Oracle and YunTable this type of relationship database What is the difference? When receiving this question, I did not answer immediately, because I feel that he is not only a question people have, and it is estimated there are a lot of students have a similar question, which is the origin of this article. But before the intervention distributed database, let's analyze what a relational database which insufficient.

 

Shortcomings of relational databases

1.  
   1. Extended difficulties: Since the presence of similar "join" statements and the like, so that a relational database difficult to extend, even can be extended, but also the more common "vertical extension" of such relatively expensive solution.
   2. Concurrent read and write slow: Because there are many internal lock relational database, resulting in a certain lack of concurrent reading and writing, when the traffic up, there may be cases performance will be significantly reduced.
   3. High cost: in addition to expensive solutions mentioned above, the acquisition cost of an ordinary commercial databases is very high.
   4. Limited support capacity: Due to similar companies like Oracle and IBM in the design of the database, not based Internet companies like Google to target customers, so that their products can meet the needs of large enterprises though, but the face of such a similar Google For massive demand, they are unsustainable.

 

 

Advantages of distributed database

1.  
   1. Simple expansion: as long as you can add a new node to a cluster through a few simple steps to support the new traffic and data, the overall size can achieve more than the P level in this regard HBase already successful examples of.
   2. Automatic backup: basic products will be distributed databases comes with automatic backup to protect data security.
   3. Fast read and write: Because memory utilization, storage, compression and Column aspects of distributed database has a lot of support, so in reading and writing, will speed advantage, examples of which are Cassandra.
   4. Low cost: Since the open-source characteristics, all of which acquisition cost is very low.

 

 

Although the relational database has many shortcomings, but there are many advantages, such as: There is good support for industry-standard SQL language, learning support and lower transaction costs, that is to say, a distributed database for substitutes database, there are many way to go, so now distributed database also belong to the minority.

Finally, notice it, YunTable immediately release its version 0.2 and 0.11, 0.11 will be a fixpack, mainly fixes memory leaks YunTable0.1 version exists, but version 0.2 is a feature-based version will be the goal full implementation of its distributed model, and be able to work properly, but does not require it can operate under the Production environment, when will be open on Google Code, I hope you have time, you can participate in together, after all, this is a very good opportunity to learn from each other.

 

Reproduced in: https: //www.cnblogs.com/licheng/archive/2010/09/09/1821912.html