canonical: https://jentic.com/apis/googleapis.com/redis

# Google Cloud Memorystore for Redis API

The Google Cloud Memorystore for Redis API creates and manages fully managed Redis instances and clusters on Google Cloud, including provisioning, resizing, version upgrades, failover, backup, import, and export operations. It exposes long-running operations for instance lifecycle changes, retrieves auth strings for client connection, and supports cluster maintenance scheduling and rescheduling. The service is a drop-in target for applications using Redis caching, session storage, and pub/sub patterns inside Google Cloud.

## For AI agents

Provision, scale, back up, and fail over fully managed Redis instances on Google Cloud, and retrieve their connection auth strings programmatically.

## Scope

Does not execute Redis data-plane commands such as GET, SET, or PUBLISH, transfer data between regions automatically, or run the Redis protocol - use for control-plane lifecycle of managed instances only.

## Capabilities

- Provision and delete managed Redis instances across regions with configurable tier and memory size
- Trigger manual failover from a primary to replica node to validate high-availability behavior
- Export an instance to a Cloud Storage RDB file and import a snapshot back into a target instance
- Schedule and reschedule cluster maintenance windows to control when patching occurs
- Retrieve the AUTH string for a Redis instance so clients can authenticate without storing static secrets
- Track long-running operations for create, update, and import flows until they complete or are cancelled

## Use cases

### Self-service Redis provisioning

Application teams request a Redis instance through an internal portal, an agent calls the Memorystore API to create it with the right tier and memory size, polls the returned long-running operation, and stores the resulting auth string in the team's secret manager. The instance is ready for use without an SRE manually clicking through the console.

Example prompt: Call POST /v1beta1/{+parent}/instances with tier=STANDARD_HA and memorySizeGb=5, poll the returned operation until done, then GET /v1beta1/{+name}/authString.

### Disaster recovery snapshots

An agent runs a nightly backup job that exports each production Redis instance to a regional Cloud Storage bucket, retains the last 14 RDB files, and validates that the export operation completed successfully. The Memorystore export endpoint produces RDB files that can be imported back into a recovery instance during incident response.

Example prompt: Iterate Redis instances under projects/my-prod, call POST /v1beta1/{+name}:export with destination=gs://backups/redis/INSTANCE-DATE.rdb on each, and verify the long-running operations complete.

### Failover testing on a schedule

Resilience engineers run a weekly chaos test: an agent calls failover on a non-production Redis instance, monitors the operation, runs a synthetic workload to confirm replica promotion, and reports the recovery time back to a metrics store. The Memorystore failover endpoint accepts a data protection mode parameter for limiting data loss tolerance.

Example prompt: Call POST /v1beta1/{+name}:failover on staging-cache with dataProtectionMode=LIMITED_DATA_LOSS and capture the elapsed time of the returned operation.

### Agent-managed cache lifecycle through Jentic

An AI agent provisions ephemeral Redis instances for short-lived test environments through Jentic. It searches for the create operation by intent, runs the call, captures the auth string, hands it to the test runner, and tears the instance down on completion. Credentials never enter the agent's context.

Example prompt: Search Jentic for 'provision google cloud redis instance', execute the create call against projects/test-env, then call delete after the test job exits.

## Key endpoints

| Method | Path | Description |
| --- | --- | --- |
| POST | /v1beta1/{+parent}/instances | Create a Redis instance |
| GET | /v1beta1/{+name} | Get a Redis instance or operation |
| PATCH | /v1beta1/{+name} | Update a Redis instance |
| DELETE | /v1beta1/{+name} | Delete a Redis instance |
| GET | /v1beta1/{+name}/authString | Retrieve the AUTH string for an instance |
| POST | /v1beta1/{+name}:failover | Trigger a failover to a replica |
| POST | /v1beta1/{+name}:export | Export an instance to Cloud Storage |
| POST | /v1beta1/{+name}:import | Import an RDB snapshot into an instance |

## Key resources

- **Instances** — Create, list, get, update, delete, export, import, failover, and back up Redis instances.
- **Clusters** — Reschedule cluster maintenance and trigger cluster-level operations.
- **Operations** — Track and cancel long-running operations produced by lifecycle calls.
- **Locations** — List the regions where Memorystore for Redis is available.

## Why Jentic

- **Setup:** Wiring the Memorystore for Redis API by hand means configuring Google OAuth 2.0, refreshing scoped access tokens, and tracking instance resource names across the control-plane and long-running operations on redis.googleapis.com. Through Jentic you install once, import the Memorystore for Redis API from the API Directory, store the OAuth credential once, and your agent calls it.
- **Permission scoping:** This API puts the instance in the URL path (/v1beta1/{+name} and /v1beta1/{+parent}/instances), so a rule can pin your agent to one Memorystore instance: it can read and patch that instance and nothing else. You choose the operations it may call, so state-changing ones like failover, import, or delete are not included unless you add them.
- **Credential handling:** Your Google OAuth credential is stored once, encrypted, by your own Jentic One instance and a scoped access token is injected at execution time. The credential never enters the agent's prompt, logs, or context.
- **Discovery method:** Agents search Jentic by intent such as 'create a Redis instance' or 'failover a Memorystore instance', and Jentic returns the matching Memorystore operation with its input schema so the agent calls the right endpoint without reading the discovery document.

## Related APIs

- **Cloud Memorystore for Memcached API** — Memcached-protocol caching alternative for workloads that do not need Redis features.
- **Cloud Spanner API** — Pairs durable relational storage with Memorystore as the read-through cache.
- **Compute Engine API** — Provisions the VPC and VM peers that connect to Memorystore instances.
- **Cloud KMS API** — Manages CMEK keys used for Memorystore instance encryption.

## FAQ

### What authentication does the Memorystore for Redis API use?

The control-plane API uses Google OAuth 2.0 with the cloud-platform scope, declared as Oauth2 and Oauth2c. Once an instance exists, the data-plane Redis connection uses the AUTH string returned from /v1beta1/{+name}/authString, which Jentic stores encrypted in its vault rather than exposing to the agent context.

### Can I run a manual Redis failover with the Memorystore API?

Yes. POST /v1beta1/{+name}:failover triggers a failover from primary to replica on STANDARD_HA tier instances, with an optional dataProtectionMode parameter of LIMITED_DATA_LOSS or FORCE_DATA_LOSS. The call returns a long-running operation that the agent can poll with GET /v1beta1/{+name}/operations until done.

### What are the rate limits for the Memorystore for Redis API?

Google Cloud applies per-project quotas to control-plane Memorystore calls; mutating operations such as create, update, and import are limited to a small number of concurrent long-running operations per project, while reads have a per-minute quota that can be raised through the Cloud Console. There is no per-call charge inside the documented quotas.

### How do I provision a Redis instance through Jentic?

Search Jentic for 'provision google cloud redis instance', load the schema for POST /v1beta1/{+parent}/instances, and execute with the desired tier and memorySizeGb. Jentic surfaces the long-running operation handle so the agent can poll for completion before retrieving the auth string.

### Is Memorystore for Redis free?

The API itself is free to call within Google Cloud quotas, but Memorystore instances are billed per gigabyte-hour by tier (BASIC or STANDARD_HA) and region. Backup exports incur Cloud Storage charges for the resulting RDB files.

### Can I import an RDB snapshot back into a Memorystore instance?

Yes. POST /v1beta1/{+name}:import accepts a Cloud Storage URI to an RDB file and replaces the contents of the target instance. The instance must be of an equal or larger memory size than the snapshot, and the operation is exposed as a long-running operation suitable for polling.

### Can I limit what my agent is allowed to do with the Google Cloud Memorystore for Redis API?

Yes. Because this API puts the instance in the URL path (/v1beta1/{+name} and /v1beta1/{+parent}/instances), a rule in your self-hosted Jentic One can pin the agent to a single Memorystore instance, letting it read and patch that instance and nothing else. You decide which operations the agent may call, so state-changing ones like failover, import, and delete are excluded unless you add them yourself. Your Google OAuth credential stays under your own control and a scoped token is injected only at execution time, so it never reaches the agent's context.