# Reading of Clustered Index Table

![`smart`](ReadingofNonclusteredIndexTable04_aa_004_20230905.png)

----------------------------------------------------
Comparison: When reading data from a Clustered Index Table, there are only two important steps involved: 1. Looking up the metadata of table T in PD; 2. Based on the metadata obtained in step 1, reading the specific region where the table is located from the 
corresponding TiKV node.
Therefore, the conclusion is that for a Clustered Index Table, since the data itself is stored in an ordered manner according to the structure of the primary key, the read operation reduces the need to read the primary key compared to a Nonclustered Index Table. Of course, if 'id' column in the 'where id = 3' condition is not the primary key column but an secondary index, then regardless of whether it is a Nonclustered Index Table or a Clustered Index Table, the main steps involved in reading records in both table modes are  determining the meta data, reading the index, and reading the table itself.