Key concepts

MemStore is a managed cache‑database service that allows the use of Redis OSS without any installation process.
Below are the primary concepts of the MemStore service.

Concept	Description
Cache Only	A caching service that stores all data in memory without using disk
Cluster	A cluster is composed of multiple shards, each with one Primary node and multiple Replica nodes – Depending on user needs, you may choose to enable or disable cluster mode, to tailor Redis configuration appropriately
Shard	A shard is the unit for partitioning the cluster dataset, made up of multiple nodes; each node plays the role of one Primary and multiple Replicas – Add shard: if the selected cluster’s load increases, a shard is added – Delete shard: if unnecessary, a shard is removed from the selected cluster
Node	A VM (Virtual Machine) with Redis OSS installed, acting as either Primary or Replica – Node type: when creating a cluster, select the appropriate node type based on data size and load – Node lifecycle and states: you can understand status based on node state values – Add node: if read load in the selected cluster increases, you add a Replica node – Delete node: if unnecessary, you delete a node from the selected cluster
High availability	Provides an environment that can continue service when a node fails – If a Primary node fails, automatically promotes the Replica with the most recent data (Automatic Failover), and then creates a new Replica to maintain the replica count (Auto Healing) – If a Replica node fails, automatically replaces it with a new node to maintain the replica count
Promotion	When high availability is used: – The user cannot request promotion; the service automatically promotes when needed When high availability is not used: – Provides a simple way for the user to manually promote a Replica node to Primary
Read/write only endpoint provided	– When cluster mode is disabled: a Primary endpoint that allows read/write and a Read‑only endpoint are provided for Redis access – When cluster mode is enabled: the Read‑only endpoint is not provided; endpoints of each node are provided instead

Cache only

In MemStore, the Redis service offered stores all data in memory without writing to disk, so you can use it solely for caching.

Cluster

A cluster is a collection of one or more nodes, automatically managed as a unit. A cluster consists of multiple shards and nodes, each serving as Primary or Replica. You can add or remove shards or Replica nodes as needed.

Cluster lifecycle and status

The lifecycle and states for a cluster are as follows.

When cluster mode enabled

info

When cluster mode is enabled, the state of shards and each node depends on the cluster status, meaning they share the same status as the cluster.

State	Description
Init	Initial state before the cluster creation starts
Creating	Creating resources
Running	Operating normally
Modifying	The cluster is being modified (adding/deleting shards, nodes)
Deleting	Deleting resources
Failed	Permanent error occurred; cannot return to normal, only deletion is possible

When cluster mode disabled

State	Description
Init	Immediately after cluster creation begins
Provisioning	All nodes in the cluster are preparing required resources
Replicating	The roles of nodes are being configured and endpoints are being created
Running	Service is fully configured and ready for access
ScalingIn	Deleting a specific Replica node
SwitchingOver	Promoting a Replica to Primary
ScalingOut	Adding a new node and assigning it as Replica
Recovering	The cluster with HA is automatically handling node failure
Modifying	Modifying cluster (e.g., changing security group)
Terminating	Deleting the cluster
Terminated	Cluster deletion completed
Warning	Temporary error occurred; can return to normal if resolved
Failed	Permanent error occurred; cannot return to normal, only deletion allowed

Shard

A shard is a group of nodes in cluster mode, partitioning the cluster data set.
The cluster can be composed of a minimum of 1 up to a maximum of 12 shards. In cluster mode, each shard can consist of minimum 2 up to maximum 5 nodes.

Node

A node is a VM (Virtual Machine) included in the cluster. When you create or delete a node, necessary operations are performed and the status automatically changes. Each node in the cluster is assigned the role of Primary or Replica. When cluster mode is disabled, you can configure up to 6 nodes in a cluster.

Role	Description
Primary	The node typically used by the user for read/write operations
Replica	Read‑only node to reduce load on the Primary

Node types

When creating a node, the user selects a type based on data size and load. More specialized node types will be supported later.

Node type	vCPU	Memory (GB)
m2a.large	2	8
m2a.xlarge	4	16
m2a.2xlarge	8	32
m2a.4xlarge	16	64
m2a.8xlarge	32	128
m2a.12xlarge	48	192
m2a.16xlarge	64	256
m2a.24xlarge	96	384
r2a.large	2	16
r2a.xlarge	4	32
r2a.2xlarge	8	64
r2a.4xlarge	16	128
r2a.8xlarge	32	256
r2a.12xlarge	48	384
r2a.16xlarge	64	512
r2a.24xlarge	96	768
t1i.small	2	2
t1i.medium	2	4
t1i.large	2	8
t1i.xlarge	4	16
t1i.2large	8	32

info

t1i instance types may take several minutes to tens of minutes to transition into Running state after initial creation.
Also, if CPU credit balance is low, workloads with high CPU usage may not function normally, so use caution.

Node lifecycle and status values

• When using cluster mode:

  • Nodes follow the same lifecycle as the cluster. See Cluster lifecycle and status.

• When cluster mode is disabled:

State	Description
Init	Each node begins creation when the cluster is being set up
Provisioning	The VM resources for service operation are being created - This step may take several minutes
Starting	After VM creation, Redis service is being installed and started
Running	Service is running normally and accessible
Modifying	Security group or other configuration change is in progress
Terminating	The user requested node deletion, and deletion is in progress
Terminated	The node has been deleted - After this, the node info is no longer accessible
Warning	Temporary error occurred; may return to normal when resolved
Failed	Permanent error occurred; cannot recover to normal state, only deletion possible

High availability

MemStore provides a high availability feature to ensure continuous service operation even when a node fails.

When cluster mode is enabled, high availability is enabled by default and cannot be changed.
When cluster mode is disabled, you can choose to enable high availability at cluster creation, and it can also be changed during operation.
You can check whether high availability is used on the cluster details tab.
Even when high availability is used, you can add or remove nodes, and any added node will automatically be included within the HA scope.

Node failures

Node failures are handled differently depending on the node role:

Role	Description
Primary node failure	If a Primary node fails, the system automatically promotes the Replica with the most recent data to Primary (Automatic Failover) and then creates a new Replica to maintain the replica count (Auto Healing)
Replica node failure	If a Replica node fails, the system automatically replaces it with a new node to maintain replica count

Promotion

When using cluster mode or high availability, users cannot request promotion; the service automatically performs promotion when needed.
When high availability is not used, manual promotion is provided: if the Primary node fails, the user can promote a Replica node to Primary.

Read/write only endpoint provided

When creating a cluster, a Read‑only endpoint is provided for client access.
This address is accessible within your selected private network (subnet) and the access permissions and security are automatically configured.

info

When cluster mode is enabled, a Read endpoint is not provided; instead, endpoints of each node are exposed.

Backup

You can back up the created cluster information for later restoration.
Currently, backup is provided first for clusters where cluster mode is not used.

Backup status values

State	Description
Available	Backup is complete and can be used for restore, copy or export
Pending	Backup request accepted but not started
Creating	Backup ongoing or being copied from original
Restoring	Being used for provisioning
Copying	Being copied to create another backup
Exporting	Being exported to user’s Object Storage
Error	Error occurred during backup
Deleting	Backup deletion in progress

Parameter application status

State	Description
Pending	Awaiting parameter application
Applying	Parameters are currently being applied to the cluster
In‑Sync	Parameters have been applied successfully to the cluster
Error‑Sync	Parameter application failed; you can retry the application (status changes to Applying on retry)
Suspended	During sequential application, if a preceding cluster failed to apply parameters, the following cluster is waiting; when status changes from Pending to Suspended, you can retry (status becomes Applying)