DB Usage Guide
The BACKND DB consists of two fields: 'read' and 'write'.
Normal read/write processing standards
When performing read/write tasks, the read throughput is calculated based on the table below.
Note: The actual throughput may change depending on the API processing method and index usage of the feature in use.
Read task
| Data Size | Throughput |
|---|---|
| Below 4 KB | 0.5 |
| 4KB | 1 |
Example
The read throughput used to load 10 KB is 1 throughput(4 KB) + 1 throughput(4 KB) + 0.5 throughput(2 KB) = 2.5 throughputs(10 KB).
Write task
| Data Size | Throughput |
|---|---|
| 1 KB or less | 1 |
Example 1
The write throughput used to save, modify and delete 96 KB is 96.
Example 2
The write throughput used when modifying 10 KB of data to 1 KB is calculated based on the larger data by comparing it before and after modification.
10 KB before modification > 1 KB after modification, therefore the write throughput is 10.
Example 3
The write throughput used when modifying 3 KB of data to 20 KB is calculated based on the larger data by comparing it before and after modification.
3 KB before modification > 20 KB after modification, therefore the write throughput is 20.
Example 4
When deleting 15 KB of data, the write throughput is the write throughput of 15.
Transaction processing standards
For transaction tasks, only up to 4 KB of data can be performed regardless of read/write tasks.
When performing transaction tasks, the read/write throughput is calculated based on the table below.
Transaction read task
| Data Size | Throughput |
|---|---|
| 4 KB or less | 2 |
Example
When loading 3.5 KB, the write throughput is 2.
Transaction write task
| Data Size | Throughput |
|---|---|
| 1 KB or less | 2 |
Example
The write throughput used to save, modify and delete 3.5 KB is 2 throughput(1 KB) + 2 throughput(1 KB) + 2 throughput(1 KB) + 2 throughput(0.5 KB) = 8 throughput.
Data type/size
When saving data to the BACKND DB, it is saved according to the saved data type.
All data saved in the BACKND DB increases by around 200 bytes per row because indexes and authentication information are added.