Shulou(Shulou.com)06/01 Report--

We all know that Hadoop is a database, but we are actually talking about Hbase. What's the difference between it and the relational database that we usually understand?

1. It is NoSQL, it does not have the interface of SQL, has its own set of API.

two。 Relational database can do summary, can do regular analysis, but Hbase can not, it can not do summary. So Hbase operation is not convenient, can not do summary, can not do analysis, what is the use? Its random reading and writing efficiency is very high, it can store a large amount of data, and it is fast to query randomly based on a certain network, a certain city and a certain machine. Or to store data based on time series, such as Wechat, Weibo, log data, it is very efficient.

3. Its storage is column, and usually the MySQL,Oracle,RDBMS we come into contact with is row storage. The difference between row storage and column storage is that row storage is suitable for online transaction scenarios and is suitable for random access, such as going to the bank to modify account records and personal information, which is to find the row where your information is located from the database. Column storage is continuous storage in column units. what would be the effect of putting column storage in the bank? If you want to modify the information, the definition to the user's row information will scan the entire table, so the row storage is mainly used for online transaction processing, while the column storage is suitable for data analysis. Because in big data's database scenario, we will build a very wide fact table, have a center through the information model, and spread many dimensions around this center. This intermediate table is a very wide table. When we conduct specific business analysis, such as analyzing the promotion of advertising business, we may only need to select some variables from a lot of data for analysis. If we use row storage, Then it is a full table scan, and using column storage will select specific parts, which is very efficient.

HBase vs RDBMS

From the above description, let's analyze the characteristics of hbase:

(1) storing massive data: Pb+

(2) High throughput: thousands of writes per node per second

(3) suitable for dealing with sparse data (semi-structured data): there is no space waste for storing empty columns of a row. Because semi-structured data has a lot of empty, then using structured data to store in a relational database, there will be a lot of space waste, and is not suitable for analysis.

But the hbase access mode is restricted, which optimizes row key-based lookups rather than full-text queries; there are no transactions and only single-line operations are supported.

Having said so much, why do we use Hbase? Here's a comparison:

(1) use HDFS

You just need to append to the dataset (no random writes)

Usually read the entire dataset (no random reads)

(2) use HBase

You need to write or read randomly.

Perform thousands of operations on TB-level data per second

(3) use RDBMS

The data is placed on a large node

Requires full transaction support

Requires the ability to query in real time

This is some personal understanding and understanding of Hbase, written and shared with you, hoping to make progress together. However, I usually follow the CSDN Forum and the official Wechat accounts such as "big data cn" and "big data Times Learning Center". From which I have learned and learned a lot, and have also taken my technical skills to a higher level. I recommend you to have a look.

Welcome to subscribe "Shulou Technology Information " to get latest news, interesting things and hot topics in the IT industry, and controls the hottest and latest Internet news, technology news and IT industry trends.