RDS 只读副本与多可用区

Databases are crucial for almost any application! But they can become slow under heavy load or go down, causing big problems for your users.

AWS offers powerful solutions to keep your databases fast and always available, even if something goes wrong.

A Read Replica is an exact copy of your primary database instance. It's designed to handle read-only traffic, taking the load off your main database.

• Perfect for read-heavy applications like blogs, news sites, or analytics dashboards.
• Supports popular engines like MySQL, PostgreSQL, MariaDB, Oracle, and SQL Server.

Read Replicas use asynchronous replication. This means changes from the primary database are copied to the replica, but not immediately. There might be a small delay, known as 'replication lag'.

Your application can then direct read queries to the replica(s) instead of the primary, improving overall performance.

• Performance Boost: Offload read queries from your primary DB, improving its write performance and overall responsiveness.
• Scalability: You can create multiple read replicas to scale your read capacity horizontally across many instances.
• Analytics: Run complex reports or analytics queries on a dedicated replica without impacting your main application's performance.

Setting up a Read Replica in AWS RDS is straightforward:

1. Navigate to your primary DB instance in the RDS console.
2. Select "Actions" and then "Create read replica".
3. Choose the instance type, storage, and other settings for your replica.
4. AWS will provision it and begin synchronizing data.

Consider a popular e-commerce website:

• Product Catalog: Users browsing products can query a Read Replica.
• User Reviews: Displaying customer reviews can also be handled by replicas.
• Reporting: Generate daily sales reports or inventory analyses on a dedicated replica.

Meanwhile, the primary database efficiently handles critical operations like new orders and inventory updates.

Multi-AZ (Multi-Availability Zone) deployments are designed for high availability and disaster recovery. When enabled, AWS automatically provisions and maintains a synchronous standby replica of your primary database in a different Availability Zone (AZ).

An AZ is a physically isolated location within an AWS Region.

Unlike Read Replicas, Multi-AZ uses synchronous replication. This means data is written to both the primary and standby instances at the same time, ensuring no data loss during a failover.

If the primary instance fails due to an outage, RDS automatically switches to the standby. Your application's database endpoint remains the same, minimizing downtime.

• Purpose: Multi-AZ is for High Availability and disaster recovery. Read Replicas are for Read Scaling.
• Replication: Multi-AZ uses synchronous replication. Read Replicas use asynchronous replication.
• Cost: Multi-AZ typically costs more due to the synchronous standby instance. Read Replicas add cost per replica.

You can use both! A Multi-AZ deployment ensures your primary database is highly available and fault-tolerant. Then, you can add one or more Read Replicas (potentially even in other AZs) to handle read scaling.

This powerful combination gives you both strong fault tolerance and excellent read performance for your applications.

You have an RDS database. Your application experiences slow read queries during peak hours, and you also want to ensure your database can survive an Availability Zone outage.

We've explored two powerful RDS features that enhance database performance and reliability:

• Read Replicas: Used for scaling read operations and improving performance for read-heavy workloads through asynchronous replication.
• Multi-AZ: Provides high availability and disaster recovery with a synchronous standby replica in another AZ and automatic failover.

Understanding when to use each is key to building robust and resilient database architectures on AWS!