Memorystore (Redis, Valkey, Memcached) Complete Guide (2026)

Memorystore is Google Cloud's in-memory data store service, offering managed Redis, Memcached, and Valkey engines. It is widely used for session stores, cache layers, real-time leaderboards, and lightweight Pub/Sub workloads.

Three Engines

Engine	Use Case	Characteristics
Redis	Mainstream, versatile	String / List / Hash / Set / ZSet / Stream / Pub/Sub
Memcached	Simple cache	Strings only, ultra-low latency
Valkey (2024+)	Redis alternative	Redis 7.2-compatible OSS, BSD license

Redis Tier Comparison

Tier	HA	Capacity	SLA	Use Case
Basic	None	1-300 GB	—	Dev, testing
Standard	Primary + Replica	1-300 GB	99.9%	Production
Cluster (2024)	Shards + Replica	1 GB to TB-scale	99.99%	Large-scale

Pricing Examples (asia-northeast1, 2026)

Configuration	Capacity	Monthly Estimate
Redis Basic	1 GB	~$35
Redis Standard	1 GB	~$70
Redis Cluster	1 shard 1 GB	~$100
Memcached	1 GB	~$30

Connection Methods

VPC: Private Service Access is required (no public IP)
GCE: Direct connection from VMs in the same VPC
GKE: Via VPC connector
Cloud Run / Functions: Serverless VPC Access or Direct VPC Egress
App Engine Standard: Via VPC connector

Key Features

AUTH: Password authentication
In-transit Encryption: TLS 1.2/1.3
At-rest Encryption: Default plus CMEK support
Persistence: RDB snapshots (1-24h), AOF
Read Replicas: Up to 5 (Standard)
Cross-region Replication: Cluster supported
Monitoring: Cloud Monitoring integration with automatic metric collection
Maintenance Window: Specify the time window for automatic upgrades

Comparison with Other Clouds

Item	Memorystore	AWS ElastiCache	Azure Cache for Redis
Engines	Redis / Memcached / Valkey	Redis / Memcached / Valkey	Redis
Cluster Mode	Yes (2024)	Yes	Yes (Premium)
Serverless	—	Yes (ElastiCache Serverless)	—
SLA	99.9-99.99%	99.9%	99.9-99.99%
Pricing Level	Medium	Medium	Medium

Typical Use Cases

Session store: Login sessions for web apps
Cache layer: Front-end DB cache to accelerate reads
Leaderboards: Real-time ranking with Sorted Sets
Rate limiting: Counters for API request rates
Lightweight Pub/Sub: Chat notifications and similar workloads
Distributed locks: Redlock pattern
ML feature cache: Reduce inference latency

Best Practices

Use Standard or higher in production to ensure HA
Manage TTL with maxmemory-policy values like allkeys-lru or volatile-lru
Monitor memory, connections, and hit rate with Cloud Monitoring
Schedule Maintenance Windows outside business hours
Use CMEK plus in-transit encryption for layered encryption

Redis or Memcached: which should I choose?

If you need persistence, complex data structures, Pub/Sub, or Streams, choose Redis. If you only need a pure cache with ultra-low latency, choose Memcached. Redis is by far the more popular choice.

What is Memorystore for Redis Cluster?

Released in 2024, this is the cluster-mode version supporting up to 250 shards, terabyte-scale memory, and a 99.99% SLA. Equivalent to AWS ElastiCache Cluster Mode.

What is Memorystore for Valkey?

Released in late 2024. Google offers a managed version of Valkey, the OSS fork of Redis. It is Redis 7.2-compatible and offers more permissive licensing.

How is pricing structured?

Memory capacity (GB) x time, multiplied by Tier (Basic / Standard / Cluster). Starts at 1 GB minimum; Standard includes the HA replica.

What are the differences between Tiers?

Basic = single node (no HA, dev use). Standard = Primary + Replica for HA. Cluster = shard-distributed for horizontal scale. Use Standard or higher in production.

Is VPC connectivity required?

Yes, required. Access goes through Private Service Access. Only private IPs are exposed; no public endpoints. You can connect from GCE, GKE, Cloud Run, and App Engine.

How does it compare to AWS ElastiCache and Azure Cache for Redis?

Feature parity is comparable, and pricing is in the same range. Memorystore is weaker on serverless options but differentiates with native Valkey support.

How does persistence work?

Memorystore for Redis supports RDB snapshots and AOF. However, note that it is designed as a cache, not as a fully persistent database.

Check what you learned with practice questions

Practice with certification-focused question sets

View GCP exam prep page

Author

NicheeLab Editorial Team

NicheeLab editorial team focused on data engineering and cloud certification learning. Content is structured around practical study needs and official exam domains.

Memorystore Complete Guide: Redis, Memcached, Valkey & Cluster Mode