Migrate to Azure Managed Redis (preview)

This article describes a number of approaches to migrate an existing Redis cache running on-premises or in another cloud service to Azure Managed Redis (preview).

Migration scenarios

Open-source Redis can run in many compute environments. Common examples include:

  • On-premises - Redis caches running in private data centers.
  • Cloud-based VMs - Redis caches running on Azure VMs, AWS EC2, and so on.
  • Hosting services - Managed Redis services such as AWS ElastiCache.

If you have such a cache, you may be able to move it to Azure Managed Redis with minimal interruption or downtime.

If you're looking to move from one Azure region to another, we recommend you see our Move Azure Managed Redis instances to different regions article.

Migration options

There are different ways that you can switch from one cache to another. Depending on where your cache is and how your application interacts with it, one method will be more useful than the others. Some of the frequently used migration strategies are detailed below.

Option Advantages Disadvantages
Create a new cache Simplest to implement. Need to repopulate data to the new cache, which may not work with many applications.
Export and import data via RDB file Compatible with any Redis cache generally. Some data could be lost, if they're written to the existing cache after the RDB file is generated.
Dual-write data to two caches No data loss or downtime. Uninterrupted operations of the existing cache. Easier testing of the new cache. Needs two caches for an extended period of time.
Migrate data programmatically Full control over how data are moved. Requires custom code.

Create a new Azure Managed Redis

This approach technically isn't a migration. If losing data isn't a concern, the easiest way to move to Azure Managed Redis is to create cache instance and connect your application to it. For example, if you use Redis as a look-aside cache of database records, you can easily rebuild the cache from scratch.

General steps to implement this option are:

  1. Create a new Azure Managed Redis instance.

  2. Update your application to use the new instance.

  3. Delete the old Redis instance.

Export data to an RDB file and import it into Azure Managed Redis

Open-source Redis defines a standard mechanism for taking a snapshot of a cache's in-memory dataset and saving it to a file. This file, called RDB, can be read by another Redis cache. You can use an RDB file to transfer data from an existing cache to Azure Managed Redis.

Important

RDB file format can change between Redis versions and may not maintain backward-compatibility. The Redis version of the cache you're exporting from should be equal or less than the version provided by Azure Managed Redis.

General steps to implement this option are:

  1. Create a new Azure Managed Redis instance in the premium tier that is the same size as (or bigger than) the existing cache.

  2. Save a snapshot of the existing Redis cache. You can configure Redis to save snapshots periodically, or run the process manually using the SAVE or BGSAVE commands. The RDB file is named “dump.rdb” by default and will be located at the path specified in the redis.conf configuration file.

    Note

    If you’re migrating data within Azure Managed Redis, see these instructions on how to export an RDB file or use the PowerShell Export cmdlet instead.

  3. Copy the RDB file to an Azure storage account in the region where your new cache is located. You can use AzCopy for this task.

  4. Import the RDB file into the new cache using these import instructions or the PowerShell Import cmdlet.

  5. Update your application to use the new cache instance.

Write to two Redis caches simultaneously during migration period

Rather than moving data directly between caches, you may use your application to write data to both an existing cache and a new one you're setting up. The application will still read data from the existing cache initially. When the new cache has the necessary data, you switch the application to that cache and retire the old one. Let's say, for example, you use Redis as a session store and the application sessions are valid for seven days. After writing to the two caches for a week, you'll be certain the new cache contains all non-expired session information. You can safely rely on it from that point onward without concern over data loss.

General steps to implement this option are:

  1. Create a new Azure Managed Redis instance in the premium tier that is the same size as (or bigger than) the existing cache.

  2. Modify application code to write to both the new and the original instances.

  3. Continue reading data from the original instance until the new instance is sufficiently populated with data.

  4. Update the application code to reading and writing from the new instance only.

  5. Delete the original instance.

Migrate programmatically

You can create a custom migration process by programmatically reading data from an existing cache and writing them into Azure Managed Redis. This open-source tool can be used to copy data from one Azure Managed Redis instance to another. A compiled version is available as well. You may also find the source code to be a useful guide for writing your own migration tool.

Note

This tool isn't officially supported by Microsoft.

General steps to implement this option are:

  1. Create a VM in the region where the existing cache is located. If your dataset is large, choose a relatively powerful VM to reduce copying time.

  2. Create a new Azure Managed Redis instance.

  3. Flush data from the new cache to ensure that it's empty. This step is required because the copy tool itself doesn't overwrite any existing key in the target cache.

    Important

    Make sure to NOT flush from the source cache.

  4. Use an application such as the open-source tool above to automate the copying of data from the source cache to the target. Remember that the copy process could take a while to complete depending on the size of your dataset.

Next steps