Working with DB instance read replicas - Amazon Relational Database Service
Services or capabilities described in Amazon Web Services documentation might vary by Region. To see the differences applicable to the China Regions, see Getting Started with Amazon Web Services in China (PDF).

Working with DB instance read replicas

A read replica is a read-only copy of a DB instance. You can reduce the load on your primary DB instance by routing queries from your applications to the read replica. In this way, you can elastically scale out beyond the capacity constraints of a single DB instance for read-heavy database workloads.

To create a read replica from a source DB instance, Amazon RDS uses the built-in replication features of the DB engine. For information about using read replicas with a specific engine, see the following sections:

After you create a read replica from a source DB instance, the source becomes the primary DB instance. When you make updates to the primary DB instance, Amazon RDS copies them asynchronously to the read replica. The following diagram shows a source DB instance replicating to a read replica in a different Availability Zone (AZ). Clients have read/write access to the primary DB instance and read-only access to the replica.

Read replica configuration

Read replicas are billed as standard DB instances at the same rates as the DB instance class used for the replica. You aren't charged for the data transfer incurred in replicating data between the source DB instance and a read replica within the same Amazon Web Services Region. For more information, see Cross-Region replication costs and DB instance billing for Amazon RDS.

Overview of Amazon RDS read replicas

The following sections discuss DB instance read replicas. For information about Multi-AZ DB cluster read replicas, see Working with Multi-AZ DB cluster read replicas for Amazon RDS.

Use cases for read replicas

Deploying one or more read replicas for a given source DB instance might make sense in a variety of scenarios, including the following:

  • Scaling beyond the compute or I/O capacity of a single DB instance for read-heavy database workloads. You can direct this excess read traffic to one or more read replicas.

  • Serving read traffic while the source DB instance is unavailable. In some cases, your source DB instance might not be able to take I/O requests, for example due to I/O suspension for backups or scheduled maintenance. In these cases, you can direct read traffic to your read replicas. For this use case, keep in mind that the data on the read replica might be "stale" because the source DB instance is unavailable.

  • Business reporting or data warehousing scenarios where you might want business reporting queries to run against a read replica, rather than your production DB instance.

  • Implementing disaster recovery. You can promote a read replica to a standalone instance as a disaster recovery solution if the primary DB instance fails.

How read replicas work

When you create a read replica, you first specify an existing DB instance as the source. Then Amazon RDS takes a snapshot of the source instance and creates a read-only instance from the snapshot. Amazon RDS then uses the asynchronous replication method for the DB engine to update the read replica whenever there is a change to the primary DB instance.

The read replica operates as a DB instance that allows only read-only connections. An exception is the RDS for Oracle DB engine, which supports replica databases in mounted mode. A mounted replica doesn't accept user connections and so can't serve a read-only workload. The primary use for mounted replicas is cross-Region disaster recovery. For more information, see Working with read replicas for Amazon RDS for Oracle.

Applications connect to a read replica just as they do to any DB instance. Amazon RDS replicates all databases from the source DB instance.

Read replicas in a Multi-AZ deployment

You can configure a read replica for a DB instance that also has a standby replica configured for high availability in a Multi-AZ deployment. Replication with the standby replica is synchronous. Unlike a read replica, a standby replica can't serve read traffic.

In the following scenario, clients have read/write access to a primary DB instance in one AZ. The primary instance copies updates asynchronously to a read replica in a second AZ and also copies them synchronously to a standby replica in a third AZ. Clients have read access only to the read replica.

Read replica and standby replica configuration

For more information about high availability and standby replicas, see Configuring and managing a Multi-AZ deployment for Amazon RDS.

Cross-Region read replicas

In some cases, a read replica resides in a different Amazon Web Services Region from its primary DB instance. In these cases, Amazon RDS sets up a secure communications channel between the primary DB instance and the read replica. Amazon RDS establishes any Amazon security configurations needed to enable the secure channel, such as adding security group entries. For more information about cross-Region read replicas, see Creating a read replica in a different Amazon Web Services Region.

The information in this chapter applies to creating Amazon RDS read replicas either in the same Amazon Web Services Region as the source DB instance, or in a separate Amazon Web Services Region. The following information doesn't apply to setting up replication with an instance that is running on an Amazon EC2 instance or that is on-premises.

Differences among read replicas for DB engines

Because Amazon RDS DB engines implement replication differently, there are several significant differences you should know about, as shown in the following table.

Feature or behavior MySQL and MariaDB Oracle PostgreSQL SQL Server

What is the replication method?

Logical replication.

Physical replication.

Physical replication.

Physical replication.

How are transaction logs purged?

RDS for MySQL and RDS for MariaDB keep any binary logs that haven't been applied.

If a primary DB instance has no cross-Region read replicas, Amazon RDS for Oracle keeps a minimum of two hours of transaction logs on the source DB instance. Logs are purged from the source DB instance after two hours or after the archive log retention hours setting has passed, whichever is longer. Logs are purged from the read replica after the archive log retention hours setting has passed only if they have been successfully applied to the database.

In some cases, a primary DB instance might have one or more cross-Region read replicas. If so, Amazon RDS for Oracle keeps the transaction logs on the source DB instance until they have been transmitted and applied to all cross-Region read replicas.

For information about setting archive log retention hours, see Retaining archived redo logs.

PostgreSQL has the parameter wal_keep_segments that dictates how many write ahead log (WAL) files are kept to provide data to the read replicas. The parameter value specifies the number of logs to keep.

The Virtual Log File (VLF) of the transaction log file on the primary replica can be truncated after it is no longer required for the secondary replicas.

The VLF can only be marked as inactive when the log records have been hardened in the replicas. Regardless of how fast the disk subsystems are in the primary replica, the transaction log will keep the VLFs until the slowest replica has hardened it.

Can a replica be made writable?

Yes. You can enable the MySQL or MariaDB read replica to be writable.

No. An Oracle read replica is a physical copy, and Oracle doesn't allow for writes in a read replica. You can promote the read replica to make it writable. The promoted read replica has the replicated data to the point when the request was made to promote it.

No. A PostgreSQL read replica is a physical copy, and PostgreSQL doesn't allow for a read replica to be made writable.

No. A SQL Server read replica is a physical copy and also doesn't allow for writes. You can promote the read replica to make it writable. The promoted read replica has the replicated data up to the point when the request was made to promote it.

Can backups be performed on the replica?

Yes. Automatic backups and manual snapshots are supported on RDS for MySQL or RDS for MariaDB read replicas.

Yes. Automatic backups and manual snapshots are supported on RDS for Oracle read replicas.

Yes, you can create a manual snapshot of RDS for PostgreSQL read replicas. Automated backups for read replicas are supported for RDS for PostgreSQL 14.1 and higher versions only. You can't turn on automated backups for PostgreSQL read replicas for RDS for PostgreSQL versions earlier than 14.1. For RDS for PostgreSQL 13 and earlier versions, create a snapshot from a read replica if you want a backup of it.

No. Automatic backups and manual snapshots aren't supported on RDS for SQL Server read replicas.

Can you use parallel replication?

Yes. All supported MariaDB and MySQL versions allow for parallel replication threads.

Yes. Redo log data is always transmitted in parallel from the primary database to all of its read replicas.

No. PostgreSQL has a single process handling replication.

Yes. Redo log data is always transmitted in parallel from the primary database to all of its read replicas.

Can you maintain a replica in a mounted rather than a read-only state?

No.

Yes. The primary use for mounted replicas is cross-Region disaster recovery. An Active Data Guard license isn't required for mounted replicas. For more information, see Working with read replicas for Amazon RDS for Oracle.

No.

No.

Read replica storage types

By default, a read replica is created with the same storage type as the source DB instance. However, you can create a read replica that has a different storage type from the source DB instance based on the options listed in the following table.

Source DB instance storage type Source DB instance storage allocation Read replica storage type options
Provisioned IOPS 100 GiB–64 TiB Provisioned IOPS, General Purpose, Magnetic
General Purpose 100 GiB–64 TiB Provisioned IOPS, General Purpose, Magnetic
General Purpose <100 GiB General Purpose, Magnetic
Magnetic 100 GiB–6 TiB Provisioned IOPS, General Purpose, Magnetic
Magnetic <100 GiB General Purpose, Magnetic
Note

When you increase the allocated storage of a read replica, it must be by at least 10 percent. If you try to increase the value by less than 10 percent, you get an error.

Restrictions for creating a replica from a replica

Amazon RDS doesn't support circular replication. You can't configure a DB instance to serve as a replication source for an existing DB instance. You can only create a new read replica from an existing DB instance. For example, if MySourceDBInstance replicates to ReadReplica1, you can't configure ReadReplica1 to replicate back to MySourceDBInstance.

For RDS for MariaDB and RDS for MySQL, and for certain versions of RDS for PostgreSQL, you can create a read replica from an existing read replica. For example, you can create new read replica ReadReplica2 from existing replica ReadReplica1. For RDS for Oracle and RDS for SQL Server, you can't create a read replica from an existing read replica.

Considerations when deleting replicas

If you no longer need read replicas, you can explicitly delete them using the same mechanisms for deleting a DB instance. If you delete a source DB instance without deleting its read replicas in the same Amazon Web Services Region, each read replica is promoted to a standalone DB instance. For information about deleting a DB instance, see Deleting a DB instance. For information about read replica promotion, see Promoting a read replica to be a standalone DB instance.

If you have cross-Region read replicas, see Cross-Region replication considerations for information related to deleting the source DB instance for a cross-Region read replica.