Restrictions on Nosql

 

650) this. width = 650; "border =" 0 "alt =" "src =" http://www.bkjia.com/uploads/allimg/131229/1U0143T8-0.jpg "/>

Nosql leads us into a new database world. Literally, we should not regard it as "No SQL ", it should be correctly understood as "Not only SQL", that is, we can Not only use SQL to query and process data, but also use programming languages and other technologies to manage our data. The data is diverse, like Darwin's evolution, data will evolve, from ancient Oracle to modern binary, as well as the recently popular QR code. Different data management methods are also different. This evolved from the database-free age-> hierarchical databases-> mesh databases-> relational databases-> non-relational databases-> the future of super databases, A few days ago, I watched the huge movie "Prometheus". There was a shocking scene "interstellar cloud map", as shown in. It showed the distribution of the entire universe in a limited space, every galaxy, every planet, is very vivid and clear. You know where you are. How is this information stored?

This may not be suitable for using relational databases for management. Why!

1. Data volume: If yuntu is to be implemented, the data volume stored first can be imagined. This can be installed without several oracle instances. Powerful storage devices are required.

 

2. Calculation workload: data is converted to graphs, and information and resources are calculated and displayed. This is not just a hardware append. It will touch the ceiling sooner or later.

 

3. transfer Traffic: if the data is non-flowing, it can only be called a specimen. to calculate the location and trajectory of the planet, in addition, to let other celestial bodies know the location of each planet and their own location, this input/output data volume can be imagined as huge.

 

4. storage Method: This yuntu can no longer simply use table to store data, because it is a non-structured data, the simplest understanding is a three-dimensional data, A table can only represent two-dimensional data. In addition, adding time is Four-Dimensional Data.

 

If you want to design a qualified Nosql database, what do you need to focus on?

1. solving data volume: I don't want to mention that distributed storage may be a good method at present, because centralized storage will touch the ceiling one day, whether it is space or capital or I/O. However, distributed storage can reduce costs, and the most important thing is to greatly improve I/O performance.

 

2. solving computational workload: this solution can adopt the cluster operation mode, that is, the human and sea tactics. The most important thing is how to coordinate the orderly work of everyone. A central control system is required to coordinate cluster work.

 

3. Solve the transmission traffic: super black cable can be used, such as black cable speed 40 Gb/s, low latency, high bandwidth. There is an immature idea, derived from the Exadata RAC architecture, which uses a whole back board integrated circuit board) to bring the two machines back and back together, and the data goes on the bus. Can we concatenate the nodes in the cluster using the bus to basically keep up with the memory I/O efficiency.

 

4. storage Solution: yuntu is a three-dimensional or multidimensional data set. I think this data can be quantified using mathematical coordinate methods. Each planet has its own coordinate set, several planets can form a galaxy, and then gradually splice them to build a shocking image in the plot.

 

Leonarding
2012.12.19
Tianjin & winter
Sharing Technology ~ Achieving dreams
Blog: www.leonarding.com

 

650) this. width = 650; "border =" 0 "alt =" "src =" http://www.bkjia.com/uploads/allimg/131229/1U0144J9-1.jpg "/>

 

This article is from the "leonarding Blog" Blog, please be sure to keep this source http://leonarding.blog.51cto.com/6045525/1094252