HBase entry concept

HBase

nosql （k，v）

Store data based on HDFS

storage structure

data model

• nameSpace: namespace
• table: table (need to declare column family column family)
• row、column
• cell: kv unit
• Version

table logical structure

• Each row of data has rowKey and column
  
  

Split storage

Horizontally split the Region and place it on different nodes
Vertically split the store and split the folders, one folder per store

Data is logically sparse with null values

The directory structure of an HBase table stored in an HDFS cluster is...

/hbase
     /<Table>             (集群中的表)
          /<Region>           (表的Regions)
               /<ColumnFamiy>      (表Regions的 ColumnFamilies)
                    /<StoreFile>        (表Regions的ColumnFamilies的 StoreFiles)

table physical structure

A line of row, the actual physical storage is a cell (kv)
row key + column family + column qualifier + timestamp + type, vlaue

In the logical view, a table can be seen as a collection of sparse rows
Physically, it is stored in a column family

HBase is based on HDFS storage, which can only read and write sequentially and append records.Cannot modify previous data,
so the update and delete operations need to be specially implemented by appending (Timestamp, Type)

Architecture todo

master
RegionServer

Read and write process

use

DDL

Create a nameSpace

Create and delete table

Add, modify, delete column family

DML

Get, Put, Scan, 和 Delete

Get can only query data based on RowKey

phoenix