Inside Xiaomi Car Interviews: SpringBoot, MySQL Tuning & Distributed Transactions

After the launch of Xiaomi's SU7 Ultra, the author became curious about the salary packages for Xiaomi automotive software positions and collected data from OfferShow, showing entry‑level offers of 22k–24k RMB per month (15‑month total, about 330k–360k RMB annually) in Beijing.

SpringBoot Startup Process

SpringBoot simplifies configuration and embeds Tomcat, allowing rapid web application development. A typical entry point is:

<code>@SpringBootApplication
public class Application {
    public static void main(String[] args) {
        SpringApplication.run(Application.class, args);
    }
}</code>

The startup sequence includes creating a

SpringApplication

instance, initializing listeners, loading the configurable environment, creating the application context, and finally refreshing the Spring container to instantiate beans and apply auto‑configuration.

Spring IOC and AOP

IOC decouples object dependencies by letting the Spring container manage creation and injection, improving maintainability and testability. Without IOC, developers must manually instantiate dependent objects with

new

, leading to tight coupling.

<code>public class UserService {
    private UserDao userDao = new UserDao();
    public void doSomething() {
        // use userDao
        userDao.saveUser();
    }
}</code>

AOP separates cross‑cutting concerns such as logging, transaction management, and security into aspects, reducing code duplication and enhancing reuse.

<code>public void doSomething() {
    System.out.println("enter doSomething");
    // business logic
    System.out.println("exit doSomething");
}</code>

MySQL Table Design and Indexing

Choose appropriate data types (e.g.,

TINYINT

for small integers,

CHAR

for fixed‑length strings,

VARCHAR

for variable length).

Limit the number of columns to reduce storage and improve query performance.

Consider denormalization when it balances read efficiency.

Create indexes on columns frequently used in

WHERE

,

JOIN

, or

ORDER BY

clauses, avoiding over‑indexing.

MySQL Lock Types

Lock Type

Scope

Statement

Description

Global Lock

Entire database

flush tables with read lock

Database becomes read‑only; all write operations block.

Table Lock

Single table

lock tables

Prevents other sessions from reading or writing the locked table.

Metadata Lock (MDL)

Table

Automatic

Read lock for CRUD, write lock for DDL to protect schema changes.

Intent Lock

Table

Automatic on DML

Marks intention to acquire row‑level locks, speeding lock checks.

Row Lock

Single record

InnoDB automatic

S (shared) and X (exclusive) locks ensure read/write mutual exclusion.

Gap Lock

Record gap

InnoDB automatic (REPEATABLE READ)

Prevents phantom reads.

Next‑Key Lock

Record + gap

InnoDB automatic

Combination of record and gap lock.

Common Design Patterns

Singleton – ensures a single instance.

Factory – abstracts object creation.

Strategy – encapsulates interchangeable algorithms.

Decorator – adds responsibilities dynamically.

Proxy – controls access to a target object, often adding logging or security.

Network Device Differences

Switches operate at the data‑link layer, forwarding frames based on MAC addresses.

Routers operate at the network layer, forwarding packets based on IP addresses using routing tables.

Distributed Transaction Solutions

Solution

Consistency

Performance

Complexity

Use Cases

2PC

Strong

Low

Medium

Traditional DB, XA

3PC

Strong

Medium‑Low

High

High‑concurrency, less strict consistency

TCC

Eventual

High

High

High‑throughput e‑commerce

Saga

Eventual

Medium

High

Long‑running business processes

Message Queue

Eventual

High

Medium

Event‑driven architectures

Local Message Table

Eventual

Medium

Low

Async notifications

Seata Framework Overview

AT mode – default, uses automatic rollback logs for relational databases.

TCC mode – requires explicit Try/Confirm/Cancel methods.

SAGA mode – splits long transactions into short ones with compensating actions.

For further learning, the author recommends the open‑source mall project (SpringBoot 3 + Vue) and its microservice variant mall‑swarm , both accompanied by video tutorials.