durumis Memorystore for Valkey (On GCP) Review - Part 1

Google recently released (although it's been a few months) Memorystore for Valkey (hereinafter referred to as Valkey).

Actually, in a previous post,Redis License the reason is explained. Due to license changes, several companies forked the existing Redis and created Valkey. That's how I understand it.

Anyway, there's tons of information about forking online (see the post above). Let's get into a more practical review of Valkey.

First, Valkey isn't a standalone product. It needs to be accessed and used by serverless applications like Cloud Run or Functions, or VMs like GCE and GKE. (Naturally, it's a Memory DB.)

There are several ways to use networks in GCP, but Valkey uses PSC (Private Service Connect).

In this review, I'll compare the pricing and network connection methods of Memorystore for Redis and Valkey.

For Redis, you can choose one of two options: Direct Peering or PSA (Private Service Access). (There are no extra charges; you just choose.)

Valkey uses PSC, and additionally uses two forwarding rules. (Additional costs apply)

Summary

Redis: You can choose from two options.

• Direct Peering
• PSA

Valkey:

• PSC: Uses two forwarding rules. (Additional costs apply)

The minimum cost for each is as follows.

Redis:

• Region: us-central1
• Memory: 1 GB
• $35.77 / month

Valkey:

• Region: us-central1
• Shard 1 without replicas (1.4 GB)
• $23.21 / month + (2 forwarding rules: $0.01 * 720 hours * 2 = $14.4) = Total $37.61

Since most of the interconnected modules are compatible, it seems best to choose based on your performance needs... However, based on my current understanding, it seems cost-effective to use Redis for low-performance needs and upgrade to Valkey when high-volume, high-performance is required.

Performance and connection details will be discussed in the next post...