Google CloudCloud MemorystoreDatabase

`google_memcache_instance` cost estimation

A managed Memcached cluster. Priced per node by vCPU-hour and memory GB-hour, multiplied across the node count.

A google_memcache_instance is Memorystore's managed Memcached offering, used as a simple in-memory cache in front of databases or APIs. Unlike Redis, Memcached on Memorystore is priced on the raw resources each node provides: vCPUs and memory, billed per hour, times the number of nodes.

Each node's cost has two parts: a per-vCPU-hour charge and a per-GB-of-memory-hour charge. A node configured with 2 vCPUs and 4 GB runs both meters every hour. The cluster total is that per-node cost multiplied by node_count. So a 2-node cluster of 2 vCPU / 4 GB nodes bills four vCPU-hours-worth and eight GB-hours-worth of capacity each hour.

c3x reads node_count and the node_config (cpu_count and memory_size_mb) from your Terraform and prices both meters across all nodes at 730 hours/month. There's no separate request or data-transfer charge for Memcached itself, which makes the estimate clean: it's purely provisioned capacity.

Terraform example

A minimal but realistic configuration that C3X can estimate.

resource "google_memcache_instance" "cache" {
  name       = "api-cache"
  region     = "us-central1"
  node_count = 2

  node_config {
    cpu_count      = 2
    memory_size_mb = 4096
  }

  memcache_version = "MEMCACHE_1_6_15"
}

Pricing dimensions

What you actually pay for when you provision google_memcache_instance.

Dimension	Unit	What's being charged
Node vCPU hours	per vCPU-hour	Per-vCPU rate, billed for cpu_count x node_count vCPUs every hour. c3x assumes 730 hours/month. $0.0466/vCPU-hour in us-central1
Node memory hours	per GB-hour	Per-GB-of-memory rate, billed for (memory_size_mb / 1024) x node_count GB every hour. $0.009/GB-hour in us-central1

Sample C3X output

Example output from c3x estimate for the 2-node cluster above:

google_memcache_instance.cache
├─ Node vCPUs (2 x 2)        2,920  vCPU-hours    $136.07
└─ Node memory (4 GB x 2)    5,840  GB-hours       $52.56

OVERALL TOTAL                                     $188.63

Optimization tips

Common ways to reduce google_memcache_instance cost without changing the workload.

Size memory to the working set, not the peak

Proportional to over-provisioning

Memcached cost scales linearly with provisioned memory. Cache hit-rate plateaus once the hot working set fits; provisioning beyond that wastes GB-hours. Measure hit rate before adding memory.

Fewer larger nodes vs many small ones

Operational, minor cost

vCPU and memory bill the same regardless of node shape, but fewer nodes reduce connection overhead and rebalancing. Consolidate unless you need the node count for availability.

Use Memcached over Redis when you only need a cache

Replica cost avoided

If you don't need persistence, replication, or data structures, Memcached is simpler and you pay only for vCPU and memory, with no replica overhead.

FAQ

How does c3x estimate Memorystore Memcached cost?

It reads node_count and node_config (cpu_count, memory_size_mb), then prices vCPU-hours and memory-GB-hours across all nodes at 730 hours/month. There is no separate request charge.

Is Memcached cheaper than Memorystore for Redis?

Often, for pure caching. Memcached has no replica or persistence overhead, so you pay only for the vCPU and memory you provision. Redis costs more because of replication and its richer feature set.

Are there data transfer charges?

Memcached itself has no per-request or egress charge within the region. Cross-region or internet egress, if any, bills under networking, not the instance.

Does node shape (vCPU vs memory ratio) change the rate?

No. vCPU-hours and memory-GB-hours bill at fixed rates regardless of ratio. The total depends only on total provisioned vCPUs and GB across the cluster.

How do I lower the cost?

Reduce provisioned memory to match the working set and avoid over-provisioning vCPUs. Both meters scale linearly, so trimming capacity directly trims the bill.