Platform Engineering Toolkit for Valkey

The best stack for the best cache

Run Valkey with the gateway, Kubernetes control plane, hardened runtime, and production modules that platform teams need for serious scale.

Valkey
  1. 01

    One stack, one operating model

    Route traffic, govern Kubernetes clusters, and run tuned Valkey nodes without stitching together unrelated components.

  2. 02

    Built from production scars

    The stack reflects Momento's experience running mission-critical cache infrastructure for high-scale applications.

  3. 03

    Keep control in your cloud

    Deploy in your environment while preserving the performance, security, and operational defaults of a managed platform.

10M+ clients and requests per second through Valkey Router
150K+ operations per second per ARM core in the Enterprise Image
250us p99 latency target from Momento's tuned Valkey runtime
80% lower cost per GB possible with NVMe-backed tiering

Architecture

The best way to run Valkey is as a platform toolkit

Each component maps to a platform engineering concern: traffic, lifecycle governance, runtime standards, storage economics, observability, and security.

Valkey stack layers Router handles traffic, Operator controls Kubernetes lifecycle, and Enterprise Image runs optimized Valkey nodes. Valkey Router Protects clusters from connection storms, hot keys, cross-region latency, and auth. Valkey Operator Gives platform teams direct, policy-aware control over Kubernetes Valkey clusters. Valkey Image Runs the tuned, hardened, CVE-patched engine with powerful optional modules. Application-compatible endpoint Kubernetes-native governance Fast, secure data plane

Products

Three products, one coherent Valkey platform

Traffic layer

Valkey Router

Scale to millions of clients and requests per second with connection pooling, throttling, request coalescing, TLS termination, JWT auth, and a web-scale HTTP API.

Control plane

Valkey Operator

Run Valkey on any CNCF-conformant Kubernetes service using purpose-built CRDs, RBAC policy, flexible namespacing, and direct pod lifecycle control.

Data plane

Valkey Enterprise Image

Deploy the same tuned Valkey container foundation that powers Momento, with always-on secure defaults, day-0 CVE patching, and optional tiering, observability, and security modules.

Traffic visual

Router absorbs the messy edge before it reaches Valkey

Client fan-out, direct HTTP access, global round trips, hot keys, and auth all terminate in one fast layer that keeps the data plane focused on commands.

Web Mobile Services Router poolcoalesceauthorize 10M+ RPS Valkey warm pooledconnections
Operator applies policy

Control visual

Operator turns Kubernetes into a Valkey control plane

Purpose-built CRDs and RBAC encode what platform teams want: deliberate versions, isolated clusters, safer rollouts, and direct control when load gets weird.

Runtime visual

The image carries the production defaults into every node

Performance tuning, CVE patching, secure runtime choices, observability, and data tiering travel with the container instead of living in scattered runbooks.

Enterprise Image 150K+ op/s per core 250us p99 latency accelerated CVE patches Data tiering OTel metrics Security package

Operate

Built for platform teams that cannot afford surprises

The Momento stack turns Valkey from a collection of moving parts into a disciplined production system.

  1. Performance without rewrites

    Keep the Valkey API while the router absorbs traffic spikes, optimizes connections, and lets clusters focus on commands.

  2. Governance by default

    Encode versions, node shapes, rollout behavior, and policy into Kubernetes-native resources that platform teams can own.

  3. Secure, observable infrastructure

    Use hardened builds, OpenTelemetry metrics, secure runtime behavior, and storage expansion options designed for long-lived production clusters.

Extensions

Modules that make Valkey cheaper, clearer, and harder to break

The Enterprise Image is the base; modules add the platform features operators normally have to assemble themselves.

Data Tiering Module

Expand string and byte storage to local NVMe or remote sinks such as S3 Express and Postgres. Zero-copy Rust read/write paths make large-object overhead negligible.

Observability Module

Emit the same advanced OpenTelemetry metrics Momento uses for hyperscale clusters, so teams can catch queue depth, replication, eviction, and saturation issues early.

Enhanced Security Package

Run without root, strip the container to minimal runtime dependencies, and compile out high-risk surfaces such as Lua scripting to reduce exposure to RCE vectors.

Pricing

Transparent self-managed pricing

Use the layers you need, then combine them as the operating model matures.

Traffic layer

Valkey Router

$0.01 per GB of data transfer

Total bytes sent and received by the router. Priced to scale with the traffic layer, not the number of clients behind it.

Control plane

Valkey Operator

$3.50 per GB-month of managed capacity

Total configured storage for Valkey instances managed by the operator across your internal Kubernetes platform.

Runtime

Valkey Enterprise Image

$1.50 per GB-month of configured capacity

Production-ready base image with optimized Valkey engine and CVE-patched builds.

  • Data Tiering Module+$1.00
  • Observability Module+$0.50
  • Enhanced Security Package+$1.50

Add-ons are priced per GB-month of configured capacity. Discounts are available for multi-year commitments, usage volume, and multi-product combinations.

Why Momento

Everything you need for the best way to run Valkey

For application teams

A stable endpoint, standard Valkey protocol compatibility, direct HTTPS options, and fewer application-level workarounds for global access, hot keys, and auth.

For platform teams

Explicit lifecycle control, Kubernetes-native policy, hardened images, production telemetry, and the confidence that every layer was designed to work together.

FAQ

Plan the stack around your environment

Do these products need to be adopted together?

No. Each product is valuable on its own, but the strongest operating model comes from using the router, operator, and image as coordinated layers of the same Valkey platform.

Can this run in our cloud?

Yes. The stack is self-managed and designed for teams running Valkey in their own cloud or Kubernetes environment while keeping control over data, security, and capacity.

What makes the image different?

The enterprise image is tuned for performance, ships with accelerated security patching, and can add modules for data tiering, observability, and stronger runtime hardening.

What changes for application developers?

Applications continue speaking the Valkey protocol through the router. Teams can also expose secure HTTPS access without bolting on another gateway layer.