Memorystore pricing is based on the following components:

• Service tier: Determines whether the Redis instance is standalone or highly available.
• Provisioned capacity: Determines how much storage and throughput is available to the Redis instance.
• Region: Determines the location where the Redis instance is provisioned.
• Replicas: Determines how many nodes are there in a Standard Tier instance. Only applicable for instances with read replicas enabled.

Memorystore provides the following tiers of service:

• Basic Tier: Provides a standalone Redis instance. Use this tier for applications that require a simple Redis cache.
• Standard Tier: Provides a highly available Redis instance that includes automatically enabled cross-zone replication and automatic failover.

Standard Tier instances support read replicas which allow adding up to five replicas to an instance and using the replicas to scale read queries. To learn more about how enabling read replicas impacts pricing, see Instance pricing with read replicas.

Note: Currently, service tier upgrades are not supported.

When you provision a Memorystore instance, you choose its maximum storage capacity. The instance falls into a capacity tier based on its provisioned capacity. For example, a 3 GiB instance falls into the 1 to 4 GiB capacity tier. The capacity tier that the instance falls into affects the following:

• The price that you pay for the provisioned capacity. For example, if you provision a 3 GiB Basic instance, you are charged based on the M1 Basic price for the region the instance is provisioned in.
• The performance you can achieve with the Redis instance. With Redis version 6, you can achieve higher throughput with higher memory tiers. For example, for typical workload patterns you can see an increase in performance when scaling from M2 to M3 or higher tier.

Pricing depends on the service tier, provisioned capacity, and region where the instance is provisioned. You are charged for a Memorystore instance in 1-second increments, based on the instance's provisioned capacity in GiB. The table below shows the price per GiB for a full hour (3,600 seconds).

To see Basic Tier and Standard Tier pricing, select a region from the drop-down list above the following table:

Charges are incurred as follows:

• Pricing is rounded up to the nearest second.
• Your project starts incurring Memorystore charges when you create a Redis instance. Whether you use the instance or not, you are charged based on the provisioned capacity.
• You can change the capacity of a Memorystore instance at any time. If the updated capacity causes your instance to fall into a different capacity tier, you are charged based on the new capacity tier as soon as Memorystore completes the scaling operation.

Standard Tier instances support read replicas. You can create instances with read replicas with the Google Cloud console, gcloud, or the Memorystore API. This feature allows you to provision up to five replicas, and the replicas can be fully utilized for reads. Multiple replicas also help improve the redundancy and availability of the instance during failures. Read replicas are only supported on M2 and higher memory tiers. To learn more about read replicas, see Read replicas.

When you enable read replica on a Standard Tier instance, the following changes occur for billing:

• The instance is billed as a new SKU.
• Pricing depends on the provisioned capacity, number of nodes, and the region where the instance is provisioned.

The following table lists the prices for nodes when read replicas are enabled for your instance:

Charges are incurred as follows:

• Pricing is rounded up to the nearest second.
• Your project starts incurring Memorystore charges when you create a Redis instance. Whether you use the instance or not, you are charged based on the provisioned capacity.
• You can change the capacity and number of replicas of a Memorystore instance at any time. If the updated capacity causes your instance to fall into a different capacity tier, you are charged based on the new capacity tier as soon as Memorystore completes the scaling operation.

Memorystore does not bill you ingress or egress from the Memorystore instance when accessing it from a client in the same region. However, your client resource might bill you for cross zone egress traffic. For example, a Compute Engine VM bills you for cross zone egress traffic to Memorystore.

When accessing a Memorystore instance from a client in a different region, Memorystore charges you for network egress traffic from Redis instances to your client application. You are charged for the total GiB transferred from one region to the other.

To view a list of prices for network egress for Memorystore, see the following table:

¹ Oceania includes Australia, New Zealand, and surrounding Pacific Ocean islands such as Papua New Guinea and Fiji. This region excludes Hawaii.

Note: Egress to Memorystore in a different region is priced the same as VM-to-VM egress pricing. This charge is billed using the standard inter-region network SKUs.

To calculate the per-hour cost of an instance, multiply the provisioned capacity with the unit price for the corresponding capacity tier.

• Calculating a Basic Tier M2 instance: If you provision a Basic Tier 8 GiB instance, the corresponding capacity tier for 8 GiB is M2. The unit cost for a Basic Tier M2 instance in the Iowa region is $0.027 per GiB per hour. The hourly cost is 8 * $0.027, which is approximately $0.22 per hour or $160.60 per month ($0.22 * 730).
• Running and scaling a Standard Tier M3 instance: If you provision a Standard Tier 20 GiB instance in the Iowa region and run the instance for 90 minutes, you are charged $1.38 (20 GiB * $0.046 GiB per hr * 1.5 hours). If you change your Standard Tier M3 instance's provisioned capacity from 20 GiB to a 50 GiB (M4) instance, you are charged the M4 price ($0.035 GiB per hr) when the scaling operation completes.

You can also use the Memorystore Pricing Calculator to get pricing estimates for your Memorystore usage scenarios.

To calculate the per-hour cost of an instance with read replica enabled, multiply the provisioned capacity with the unit price for a node for the corresponding capacity tier and the number of nodes.

For example, if you enable read replicas you are charged $2.85(50 * $0.019 * 2 * 1.5), where 50GiB is the provisioned capacity, $0.019 is the unit price in Gib-hr for an M4 node, 2 is the number of nodes (1 primary and 1 replica) and 1.5 is the usage in hours.