Determining your requirements - Amazon MemoryDB for Redis
Memory and processor requirementsMemoryDB cluster configurationEnhanced I/O MultiplexingScaling requirementsAccess requirementsRegion and Availability Zones
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).

Determining your requirements

Preparation

Knowing the answers to the following questions helps make creating your cluster go smoother:

  • Make sure to create a subnet group in the same VPC before you start creating a cluster. Alternatively, you can use the default subnet group provided. For more information, see Subnets and subnet groups.

    MemoryDB is designed to be accessed from within Amazon using Amazon EC2. However, if you launch in a VPC based on Amazon VPC, you can provide access from outside Amazon. For more information, see Accessing MemoryDB resources from outside Amazon.

  • Do you need to customize any parameter values?

    If you do, create a custom parameter group. For more information, see Creating a parameter group.

  • Do you need to create a VPC security group?

    For more information, see Security in Your VPC.

  • How do you intend to implement fault tolerance?

    For more information, see Mitigating Failures.

Memory and processor requirements

The basic building block of MemoryDB for Redis is the node. Nodes are configured in shards to form clusters. When determining the node type to use for your cluster, take the cluster’s node configuration and the amount of data you have to store into consideration.

MemoryDB cluster configuration

MemoryDB clusters are comprised of from 1 to 500 shards. The data in a MemoryDB cluster is partitioned across the shards in the cluster. Your application connects with a MemoryDB cluster using a network address called an Endpoint. In addition to the node endpoints, the MemoryDB cluster itself has an endpoint called the cluster endpoint. Your application can use this endpoint to read from or write to the cluster, leaving the determination of which node to read from or write to up to MemoryDB.

Enhanced I/O Multiplexing

If you are running Redis version 7.0 or higher, you will get additional acceleration with enhanced I/O multiplexing, where each dedicated network IO thread pipelines commands from multiple clients into the Redis engine, taking advantage of Redis' ability to efficiently process commands in batches. For more information, see Ultra-fast performance and Supported node types.

Scaling requirements

All clusters can be scaled up a larger node type. When you scale up a MemoryDB cluster, you can do it online so the cluster remains available or you can seed a new cluster from a snapshot and avoid having the new cluster start out empty.

For more information, see Scaling in this guide.

Access requirements

By design, MemoryDB clusters are accessed from Amazon EC2 instances. Network access to a MemoryDB cluster is limited to the account that created the cluster. Therefore, before you can access a cluster from an Amazon EC2 instance, you must authorize ingress to the cluster. For detailed instructions, see Step 2: Authorize access to the cluster in this guide.

Region and Availability Zones

By locating your MemoryDB clusters in an Amazon Region close to your application you can reduce latency. If your cluster has multiple nodes, locating your nodes in different Availability Zones can reduce the impact of failures on your cluster.

For more information, see the following: