Scaling clusters in Valkey or Redis OSS (Cluster Mode Enabled)
As demand on your clusters changes, you might decide to improve performance or reduce costs by changing the number of shards in your Valkey or Redis OSS (cluster mode enabled) cluster. We recommend using online horizontal scaling to do so, because it allows your cluster to continue serving requests during the scaling process.
Conditions under which you might decide to rescale your cluster include the following:
-
Memory pressure:
If the nodes in your cluster are under memory pressure, you might decide to scale out so that you have more resources to better store data and serve requests.
You can determine whether your nodes are under memory pressure by monitoring the following metrics: FreeableMemory, SwapUsage, and BytesUseForCache.
-
CPU or network bottleneck:
If latency/throughput issues are plaguing your cluster, you might need to scale out to resolve the issues.
You can monitor your latency and throughput levels by monitoring the following metrics: CPUUtilization, NetworkBytesIn, NetworkBytesOut, CurrConnections, and NewConnections.
-
Your cluster is over-scaled:
Current demand on your cluster is such that scaling in doesn't hurt performance and reduces your costs.
You can monitor your cluster's use to determine whether or not you can safely scale in using the following metrics: FreeableMemory, SwapUsage, BytesUseForCache, CPUUtilization, NetworkBytesIn, NetworkBytesOut, CurrConnections, and NewConnections.
Performance Impact of Scaling
When you scale using the offline process, your cluster is offline for a significant portion of the process and thus unable to serve requests. When you scale using the online method, because scaling is a compute-intensive operation, there is some degradation in performance, nevertheless, your cluster continues to serve requests throughout the scaling operation. How much degradation you experience depends upon your normal CPU utilization and your data.
There are two ways to scale your Valkey or Redis OSS (cluster mode enabled) cluster; horizontal and vertical scaling.
-
Horizontal scaling allows you to change the number of node groups (shards) in the replication group by adding or removing node groups (shards). The online resharding process allows scaling in/out while the cluster continues serving incoming requests.
Configure the slots in your new cluster differently than they were in the old cluster. Offline method only.
-
Vertical Scaling - Change the node type to resize the cluster. The online vertical scaling allows scaling up/down while the cluster continues serving incoming requests.
If you are reducing the size and memory capacity of the cluster, by either scaling in or scaling down, ensure that the new configuration has sufficient memory for your data and Valkey or Redis OSS overhead.
For more information, see Select cache node size.
Contents
Offline resharding for Valkey or Redis OSS (cluster mode enabled)
The main advantage you get from offline shard reconfiguration is that you can do more than merely add or remove shards from your replication group. When you reshard and rebalance offline, in addition to changing the number of shards in your replication group, you can do the following:
Note
Offline resharding is not supported on Valkey or Redis OSS clusters with data tiering enabled. For more information, see Data tiering in ElastiCache.
-
Change the node type of your replication group.
-
Specify the Availability Zone for each node in the replication group.
-
Upgrade to a newer engine version.
-
Specify the number of replica nodes in each shard independently.
-
Specify the keyspace for each shard.
The main disadvantage of offline shard reconfiguration is that your cluster is offline beginning with the restore portion of the process and continuing until you update the endpoints in your application. The length of time that your cluster is offline varies with the amount of data in your cluster.
To reconfigure your shards Valkey or Redis OSS (cluster mode enabled) cluster offline
-
Create a manual backup of your existing Valkey or Redis OSS cluster. For more information, see Taking manual backups.
-
Create a new cluster by restoring from the backup. For more information, see Restoring from a backup into a new cache.
-
Update the endpoints in your application to the new cluster's endpoints. For more information, see Finding connection endpoints in ElastiCache.
Online resharding for Valkey or Redis OSS (cluster mode enabled)
By using online resharding and shard rebalancing with ElastiCache Valkey 7.2 or newer, or Redis OSS version 3.2.10 or newer, you can scale your Valkey or Redis OSS (cluster mode enabled) cluster dynamically with no downtime. This approach means that your cluster can continue to serve requests even while scaling or rebalancing is in process.
You can do the following:
-
Scale out – Increase read and write capacity by adding shards (node groups) to your Valkey or Redis OSS (cluster mode enabled) cluster (replication group).
If you add one or more shards to your replication group, the number of nodes in each new shard is the same as the number of nodes in the smallest of the existing shards.
-
Scale in – Reduce read and write capacity, and thereby costs, by removing shards from your Valkey or Redis OSS (cluster mode enabled) cluster.
-
Rebalance – Move the keyspaces among the shards in your Valkey or Redis OSS (cluster mode enabled) cluster so they are as equally distributed among the shards as possible.
You can't do the following:
-
Configure shards independently:
You can't specify the keyspace for shards independently. To do this, you must use the offline process.
Currently, the following limitations apply to ElastiCache online resharding and rebalancing:
-
These processes require Valkey 7.2 and newer or Redis OSS engine version 3.2.10 or newer. For information on upgrading your engine version, see Version Management for ElastiCache.
-
There are limitations with slots or keyspaces and large items:
If any of the keys in a shard contain a large item, that key isn't migrated to a new shard when scaling out or rebalancing. This functionality can result in unbalanced shards.
If any of the keys in a shard contain a large item (items greater than 256 MB after serialization), that shard isn't deleted when scaling in. This functionality can result in some shards not being deleted.
-
When scaling out, the number of nodes in any new shards equals the number of nodes in the smallest existing shard.
-
When scaling out, any tags that are common to all existing shards are copied to the new shards.
-
When scaling out a Global Data Store cluster, ElastiCache will not automatically replicate Functions from one of the existing nodes to the new node(s). We recommend loading your Functions in the new shard(s) after scaling out your cluster so that every shards have the same functions.
Note
In ElastiCache with Valkey 7.2 and above, and with Redis OSS version 7 and above: When scaling out your cluster, ElastiCache will automatically replicate the Functions loaded in one of the existing nodes (selected at random) to the new node(s).
If your application uses Functions
For more information, see Online cluster resizing.
You can horizontally scale or rebalance your Valkey or Redis OSS (cluster mode enabled) clusters using the AWS Management Console, the AWS CLI, and the ElastiCache API.
Adding shards with online resharding
You can add shards to your Valkey or Redis OSS (cluster mode enabled) cluster using the AWS Management Console, AWS CLI, or ElastiCache API. When you add shards to a Valkey or Redis OSS (cluster mode enabled) cluster, any tags on the existing shards are copied over to the new shards.
Topics
You can use the AWS Management Console to add one or more shards to your Valkey or Redis OSS (cluster mode enabled) cluster. The following procedure describes the process.
To add shards to your Valkey or Redis OSS (cluster mode enabled) cluster
-
Open the ElastiCache console at https://console.aws.amazon.com/elasticache/
. -
From the navigation pane, choose Valkey clusters or Redis OSS clusters.
-
Locate and choose the name, not the box to the left of the cluster's name, of the Valkey or Redis OSS (cluster mode enabled) cluster that you want to add shards to.
Tip
Valkey or Redis OSS (cluster mode enabled) show Clustered Valkey or Clustered Redis OSS in the Mode column
-
Choose Add shard.
-
For Number of shards to be added, choose the number of shards you want added to this cluster.
-
For Availability zone(s), choose either No preference or Specify availability zones.
-
If you chose Specify availability zones, for each node in each shard, select the node's Availability Zone from the list of Availability Zones.
-
Choose Add.
-
The following process describes how to reconfigure the shards in your Valkey or Redis OSS (cluster mode enabled) cluster by adding shards using the AWS CLI.
Use the following parameters with modify-replication-group-shard-configuration
.
Parameters
-
--apply-immediately
– Required. Specifies the shard reconfiguration operation is to be started immediately. -
--replication-group-id
– Required. Specifies which replication group (cluster) the shard reconfiguration operation is to be performed on. -
--node-group-count
– Required. Specifies the number of shards (node groups) to exist when the operation is completed. When adding shards, the value of--node-group-count
must be greater than the current number of shards.Optionally, you can specify the Availability Zone for each node in the replication group using
--resharding-configuration
. -
--resharding-configuration
– Optional. A list of preferred Availability Zones for each node in each shard in the replication group. Use this parameter only if the value of--node-group-count
is greater than the current number of shards. If this parameter is omitted when adding shards, Amazon ElastiCache selects the Availability Zones for the new nodes.
The following example reconfigures the keyspaces over four shards in a Valkey or Redis OSS (cluster mode enabled)
cluster named my-cluster
. The example also specifies the Availability
Zone for each node in each shard. The operation begins immediately.
Example - Adding Shards
For Linux, macOS, or Unix:
aws elasticache modify-replication-group-shard-configuration \ --replication-group-id
my-cluster
\ --node-group-count4
\ --resharding-configuration \ "PreferredAvailabilityZones=us-east-2a
,us-east-2c
" \ "PreferredAvailabilityZones=us-east-2b
,us-east-2a
" \ "PreferredAvailabilityZones=us-east-2c
,us-east-2d
" \ "PreferredAvailabilityZones=us-east-2d
,us-east-2c
" \ --apply-immediately
For Windows:
aws elasticache modify-replication-group-shard-configuration ^ --replication-group-id
my-cluster
^ --node-group-count4
^ --resharding-configuration ^ "PreferredAvailabilityZones=us-east-2a
,us-east-2c
" ^ "PreferredAvailabilityZones=us-east-2b
,us-east-2a
" ^ "PreferredAvailabilityZones=us-east-2c
,us-east-2d
" ^ "PreferredAvailabilityZones=us-east-2d
,us-east-2c
" ^ --apply-immediately
For more information, see modify-replication-group-shard-configuration in the AWS CLI documentation.
You can use the ElastiCache API to reconfigure the shards in your Valkey or Redis OSS (cluster mode enabled) cluster
online by using the ModifyReplicationGroupShardConfiguration
operation.
Use the following parameters with ModifyReplicationGroupShardConfiguration
.
Parameters
-
ApplyImmediately=true
– Required. Specifies the shard reconfiguration operation is to be started immediately. -
ReplicationGroupId
– Required. Specifies which replication group (cluster) the shard reconfiguration operation is to be performed on. -
NodeGroupCount
– Required. Specifies the number of shards (node groups) to exist when the operation is completed. When adding shards, the value ofNodeGroupCount
must be greater than the current number of shards.Optionally, you can specify the Availability Zone for each node in the replication group using
ReshardingConfiguration
. -
ReshardingConfiguration
– Optional. A list of preferred Availability Zones for each node in each shard in the replication group. Use this parameter only if the value ofNodeGroupCount
is greater than the current number of shards. If this parameter is omitted when adding shards, Amazon ElastiCache selects the Availability Zones for the new nodes.
The following process describes how to reconfigure the shards in your Valkey or Redis OSS (cluster mode enabled) cluster by adding shards using the ElastiCache API.
Example - Adding Shards
The following example adds node groups to the Valkey or Redis OSS (cluster mode enabled) cluster
my-cluster
, so there are a total of four node groups
when the operation completes. The example also specifies the
Availability Zone for each node in each shard. The operation begins
immediately.
https://elasticache.us-east-2.amazonaws.com/ ?Action=ModifyReplicationGroupShardConfiguration &ApplyImmediately=true &NodeGroupCount=
4
&ReplicationGroupId=my-cluster
&ReshardingConfiguration.ReshardingConfiguration.1.PreferredAvailabilityZones.AvailabilityZone.1=us-east-2a
&ReshardingConfiguration.ReshardingConfiguration.1.PreferredAvailabilityZones.AvailabilityZone.2=us-east-2c
&ReshardingConfiguration.ReshardingConfiguration.2.PreferredAvailabilityZones.AvailabilityZone.1=us-east-2b
&ReshardingConfiguration.ReshardingConfiguration.2.PreferredAvailabilityZones.AvailabilityZone.2=us-east-2a
&ReshardingConfiguration.ReshardingConfiguration.3.PreferredAvailabilityZones.AvailabilityZone.1=us-east-2c
&ReshardingConfiguration.ReshardingConfiguration.3.PreferredAvailabilityZones.AvailabilityZone.2=us-east-2d
&ReshardingConfiguration.ReshardingConfiguration.4.PreferredAvailabilityZones.AvailabilityZone.1=us-east-2d
&ReshardingConfiguration.ReshardingConfiguration.4.PreferredAvailabilityZones.AvailabilityZone.2=us-east-2c
&Version=2015-02-02 &SignatureVersion=4 &SignatureMethod=HmacSHA256 &Timestamp=20171002T192317Z &X-Amz-Credential=<credential>
For more information, see ModifyReplicationGroupShardConfiguration in the ElastiCache API Reference.
Removing shards with online resharding
You can remove shards from your Valkey or Redis OSS (cluster mode enabled) cluster using the AWS Management Console, AWS CLI, or ElastiCache API.
Removing shards (Console)
The following process describes how to reconfigure the shards in your Valkey or Redis OSS (cluster mode enabled) cluster by removing shards using the AWS Management Console.
Before removing node groups (shards) from your replication group, ElastiCache makes sure that all your data will fit in the remaining shards. If the data will fit, the specified shards are deleted from the replication group as requested. If the data won't fit in the remaining node groups, the process is terminated and the replication group is left with the same node group configuration as before the request was made.
You can use the AWS Management Console to remove one or more shards from your Valkey or Redis OSS (cluster mode enabled) cluster. You cannot remove all the shards in a replication group. Instead, you must delete the replication group. For more information, see Deleting a replication group. The following procedure describes the process for deleting one or more shards.
To remove shards from your Valkey or Redis OSS (cluster mode enabled) cluster
-
Open the ElastiCache console at https://console.aws.amazon.com/elasticache/
. -
From the navigation pane, choose Valkey clusters or Redis OSS clusters.
-
Locate and choose the name, not the box to the left of the cluster's name, of the Valkey or Redis OSS (cluster mode enabled) cluster you want to remove shards from.
Tip
Valkey or Redis OSS (cluster mode enabled) clusters have a value of 1 or greater in the Shards column.
-
From the list of shards, choose the box to the left of the name of each shard that you want to delete.
-
Choose Delete shard.
Removing shards (AWS CLI)
The following process describes how to reconfigure the shards in your Valkey or Redis OSS (cluster mode enabled) cluster by removing shards using the AWS CLI.
Important
Before removing node groups (shards) from your replication group, ElastiCache makes sure that
all your data will fit in the remaining shards. If the data will fit, the
specified shards (--node-groups-to-remove
) are deleted from
the replication group as requested and their keyspaces mapped into the
remaining shards. If the data will not fit in the remaining node
groups, the process is terminated and the replication group is left with the same node group configuration as before the request was made.
You can use the AWS CLI to remove one or more shards from your Valkey or Redis OSS (cluster mode enabled) cluster. You cannot remove all the shards in a replication group. Instead, you must delete the replication group. For more information, see Deleting a replication group.
Use the following parameters with modify-replication-group-shard-configuration
.
Parameters
-
--apply-immediately
– Required. Specifies the shard reconfiguration operation is to be started immediately. -
--replication-group-id
– Required. Specifies which replication group (cluster) the shard reconfiguration operation is to be performed on. -
--node-group-count
– Required. Specifies the number of shards (node groups) to exist when the operation is completed. When removing shards, the value of--node-group-count
must be less than the current number of shards. -
--node-groups-to-remove
– Required when--node-group-count
is less than the current number of node groups (shards). A list of shard (node group) IDs to remove from the replication group.
The following procedure describes the process for deleting one or more shards.
Example - Removing Shards
The following example removes two node groups from the Valkey or Redis OSS (cluster mode enabled) cluster
my-cluster
, so there are a total of two node groups
when the operation completes. The keyspaces from the removed shards are
distributed evenly over the remaining shards.
For Linux, macOS, or Unix:
aws elasticache modify-replication-group-shard-configuration \ --replication-group-id
my-cluster
\ --node-group-count2
\ --node-groups-to-remove "0002
" "0003
" \ --apply-immediately
For Windows:
aws elasticache modify-replication-group-shard-configuration ^ --replication-group-id
my-cluster
^ --node-group-count2
^ --node-groups-to-remove "0002
" "0003
" ^ --apply-immediately
Removing shards (ElastiCache API)
You can use the ElastiCache API to reconfigure the shards in your Valkey or Redis OSS (cluster mode enabled) cluster
online by using the ModifyReplicationGroupShardConfiguration
operation.
The following process describes how to reconfigure the shards in your Valkey or Redis OSS (cluster mode enabled) cluster by removing shards using the ElastiCache API.
Important
Before removing node groups (shards) from your replication group, ElastiCache makes sure that
all your data will fit in the remaining shards. If the data will fit, the
specified shards (NodeGroupsToRemove
) are deleted from the
replication group as requested and their keyspaces mapped into the
remaining shards. If the data will not fit in the remaining node
groups, the process is terminated and the replication group is left with the same node group configuration as before the request was made.
You can use the ElastiCache API to remove one or more shards from your Valkey or Redis OSS (cluster mode enabled) cluster. You cannot remove all the shards in a replication group. Instead, you must delete the replication group. For more information, see Deleting a replication group.
Use the following parameters with ModifyReplicationGroupShardConfiguration
.
Parameters
-
ApplyImmediately=true
– Required. Specifies the shard reconfiguration operation is to be started immediately. -
ReplicationGroupId
– Required. Specifies which replication group (cluster) the shard reconfiguration operation is to be performed on. -
NodeGroupCount
– Required. Specifies the number of shards (node groups) to exist when the operation is completed. When removing shards, the value ofNodeGroupCount
must be less than the current number of shards. -
NodeGroupsToRemove
– Required when--node-group-count
is less than the current number of node groups (shards). A list of shard (node group) IDs to remove from the replication group.
The following procedure describes the process for deleting one or more shards.
Example - Removing Shards
The following example removes two node groups from the Valkey or Redis OSS (cluster mode enabled) cluster
my-cluster
, so there are a total of two node groups
when the operation completes. The keyspaces from the removed shards are
distributed evenly over the remaining shards.
https://elasticache.us-east-2.amazonaws.com/ ?Action=ModifyReplicationGroupShardConfiguration &ApplyImmediately=true &NodeGroupCount=
2
&ReplicationGroupId=my-cluster
&NodeGroupsToRemove.member.1=0002
&NodeGroupsToRemove.member.2=0003
&Version=2015-02-02 &SignatureVersion=4 &SignatureMethod=HmacSHA256 &Timestamp=20171002T192317Z &X-Amz-Credential=<credential>
Online shard rebalancing
You can rebalance shards in your Valkey or Redis OSS (cluster mode enabled) cluster using the AWS Management Console, AWS CLI, or ElastiCache API.
Topics
Online Shard Rebalancing (Console)
The following process describes how to reconfigure the shards in your Valkey or Redis OSS (cluster mode enabled) cluster by rebalancing shards using the AWS Management Console.
To rebalance the keyspaces among the shards on your Valkey or Redis OSS (cluster mode enabled) cluster
-
Open the ElastiCache console at https://console.aws.amazon.com/elasticache/
. -
From the navigation pane, choose Valkey clusters or Redis OSS clusters.
-
Choose the name, not the box to the left of the name, of the Valkey or Redis OSS (cluster mode enabled) cluster that you want to rebalance.
Tip
Valkey or Redis OSS (cluster mode enabled) clusters have a value of 1 or greater in the Shards column.
-
Choose Rebalance.
-
When prompted, choose Rebalance. You might see a message similar to this one:
Slots in the replication group are uniformly distributed. Nothing to do. (Service: AmazonElastiCache; Status Code: 400; Error Code: InvalidReplicationGroupState; Request ID: 2246cebd-9721-11e7-8d5b-e1b0f086c8cf)
. If you do, choose Cancel.
Online shard rebalancing (AWS CLI)
Use the following parameters with modify-replication-group-shard-configuration
.
Parameters
-
-apply-immediately
– Required. Specifies the shard reconfiguration operation is to be started immediately. -
--replication-group-id
– Required. Specifies which replication group (cluster) the shard reconfiguration operation is to be performed on. -
--node-group-count
– Required. To rebalance the keyspaces across all shards in the cluster, this value must be the same as the current number of shards.
The following process describes how to reconfigure the shards in your Valkey or Redis OSS (cluster mode enabled) cluster by rebalancing shards using the AWS CLI.
Example - Rebalancing the Shards in a Cluster
The following example rebalances the slots in the Valkey or Redis OSS (cluster mode enabled) cluster my-cluster
so that the slots are distributed as equally as possible.
The value of --node-group-count
(4
) is the number of
shards currently in the cluster.
For Linux, macOS, or Unix:
aws elasticache modify-replication-group-shard-configuration \ --replication-group-id
my-cluster
\ --node-group-count4
\ --apply-immediately
For Windows:
aws elasticache modify-replication-group-shard-configuration ^ --replication-group-id
my-cluster
^ --node-group-count4
^ --apply-immediately
Online shard rebalancing (ElastiCache API)
You can use the ElastiCache API to reconfigure the shards in your Valkey or Redis OSS (cluster mode enabled) cluster
online by using the ModifyReplicationGroupShardConfiguration
operation.
Use the following parameters with ModifyReplicationGroupShardConfiguration
.
Parameters
-
ApplyImmediately=true
– Required. Specifies the shard reconfiguration operation is to be started immediately. -
ReplicationGroupId
– Required. Specifies which replication group (cluster) the shard reconfiguration operation is to be performed on. -
NodeGroupCount
– Required. To rebalance the keyspaces across all shards in the cluster, this value must be the same as the current number of shards.
The following process describes how to reconfigure the shards in your Valkey or Redis OSS (cluster mode enabled) cluster by rebalancing the shards using the ElastiCache API.
Example - Rebalancing a Cluster
The following example rebalances the slots in the Valkey or Redis OSS (cluster mode enabled) cluster my-cluster
so that the slots are distributed as equally as possible.
The value of NodeGroupCount
(4
) is the number of
shards currently in the cluster.
https://elasticache.us-east-2.amazonaws.com/ ?Action=ModifyReplicationGroupShardConfiguration &ApplyImmediately=true &NodeGroupCount=
4
&ReplicationGroupId=my-cluster
&Version=2015-02-02 &SignatureVersion=4 &SignatureMethod=HmacSHA256 &Timestamp=20171002T192317Z &X-Amz-Credential=<credential>