All Products
Search
Document Center

Container Service for Kubernetes:Overview of multi-cluster gateways

Last Updated:Mar 26, 2026

ACK One's multi-cluster gateway is a solution for application disaster recovery and north-south traffic management in hybrid cloud and multi-cluster Kubernetes environments. It helps you quickly implement zone-disaster recovery or geo-disaster recovery for hybrid cloud and multi-cluster applications. Instead of configuring ingress rules separately in each cluster, you define routing rules once on the Fleet instance and let the gateway distribute traffic across all clusters — with millisecond-level fallback when a cluster becomes unhealthy.

Two gateway types are available: ALB multi-cluster gateways and MSE (Microservices Engine) multi-cluster gateways. Each type targets different deployment topologies and protocol requirements.

Architecture

image

How it works

  1. Create an AlbConfig or MseIngressConfig resource on the ACK One Fleet instance. This provisions an ALB or MSE multi-cluster gateway.

  2. Create an Ingress on the Fleet instance and configure routing rules.

  3. The multi-cluster Ingress controller reads those rules and routes incoming traffic to backend servers across all managed clusters.

Key capabilities

  • Fully managed and O&M-free — no gateway infrastructure to operate or maintain.

  • Reduced gateway sprawl — one region-level Layer 7 gateway serves multiple clusters instead of one gateway per cluster, lowering cost.

  • Centralized routing — configure forwarding rules for all clusters in one place on the Fleet instance.

  • Cross-zone high availability — the gateway is designed for zone-level high availability (HA) from the ground up.

  • Millisecond-level fallback — if a backend server error occurs in one cluster, traffic is smoothly redirected to healthy backends with no noticeable interruption.

Choose a gateway type

Answer the following questions to identify the right gateway type:

  1. Do you need cross-region traffic management? ALB supports it; MSE does not.

  2. Are your on-premises or third-party cloud clusters using overlay networks? ALB supports both overlay and underlay networks; MSE supports underlay only.

  3. Do you need HTTP/2 or gRPCS? MSE supports HTTP/2 and gRPCS; ALB does not.

If your answer to question 1 or 2 is yes, use ALB. If your workloads are confined to the same VPC or same region and you need HTTP/2 or gRPCS, use MSE.

The following table compares the two gateway types in detail.

ALB multi-cluster gateway MSE multi-cluster gateway
Deployment scope Cross-zone, cross-region, hybrid cloud, cross-cloud Same VPC (default); cross-VPC in the same region via Cloud Enterprise Network (CEN)
Supported scenarios Active zone-disaster recovery; active geo-disaster recovery; hybrid cloud and active cross-cloud disaster recovery (underlay and overlay networks); weight-based traffic splitting; header-based routing to a specified cluster Active zone-disaster recovery or active-standby disaster recovery; hybrid cloud or cross-cloud disaster recovery (underlay networks); weight-based traffic splitting; header-based routing to a specified cluster
Network requirements Cross-region: connect networks using CEN; node and pod CIDR blocks must not overlap. On-premises or third-party cloud clusters: connect using Express Connect circuits. Same-VPC clusters: no extra setup. Cross-VPC (same region): connect using CEN; node and pod CIDR blocks must not overlap.
Limitations No cross-region restrictions. Cannot deploy across regions. On-premises clusters using overlay networks and third-party public cloud Kubernetes clusters are not supported.
Backend protocols HTTP, HTTPS, gRPC, QUIC HTTP, HTTP/2, HTTPS, gRPC, gRPCS
Billing ALB billing Billing overview of common instances

What's next

After choosing a gateway type, set up the gateway and configure traffic routing for your clusters:

  • Create an ALB multi-cluster gateway

  • Create an MSE multi-cluster gateway