Container Compute Service:Recommendations for selecting a Knative gateway

Type

ALB

ASM

Kourier

Product positioning

• Focuses on application layer payloads. ALB provides Layer 7 load balancing and is tightly integrated with container technology.

• Supports application-layer protocols such as HTTP, HTTPS, and QUIC. It is suitable for scenarios that require high capacity and automatic scaling.

• Supports multiple deployment strategies, such as canary releases, A/B testing, and blue-green deployments. It can be used with other cloud products such as WAF, FC, PrivateLink, and TR.

Provides a fully managed service mesh platform that is compatible with the open source Istio service mesh. It is designed to simplify service administration by managing traffic routing and splitting for service invocations, ensuring secure inter-service communication, and providing mesh observability. This reduces the workload for developers and O&M teams.

A lightweight gateway based on Envoy. It is the gateway implementation provided by the community for accessing Knative Serving services. It provides essential routing and service discovery features.

Service architecture

• Built on the Alibaba Cloud Apsara Cloud Network platform.

• Based on the self-developed CyberStar platform and supports automatic scaling.

• The components of the Istio control plane are fully managed. It remains compatible with the Istio community.

• A single ASM instance with a managed control plane can support application services from multiple Kubernetes clusters or application services that run on ECI pods.

• Based on Envoy.

• You can manually configure the number of replicas and resource limits.

Basic routing

• Content-based and source IP-based routing.

• HTTP header modification, redirection, rewrite, throttling, cross-domain access, and session persistence.

• Supports forwarding rules for both requests and responses.

• Custom traffic routing rules.

• Spans multiple Kubernetes clusters.

• Fine-grained traffic management.

• Out-of-the-box chaos engineering capabilities.

• Content-based routing.

• HTTP header modification.

O&M capabilities

• Fully managed and configuration-free.

• Automatic scaling and supports ultra-large capacity.

• Processing capacity automatically scales with your business traffic.

• One-click installation, deployment, and upgrades.

• Managed control plane components.

• Lets you focus on business application development.

• Compatible with Istio community specifications.

• Self-maintained components.

• You can use Horizontal Pod Autoscaler (HPA) configurations to scale in or out.

• Requires proactive configuration for performance tuning.

Performance

• A single instance supports 1 million queries per second (QPS).

• A single instance supports tens of millions of concurrent connections.

• Uses SSL hardware acceleration by default.

• Supports multi-region deployment and provides low-latency access for clients. When combined with intelligent DNS parsing, it can resolve a domain name to the IP address of the geographically closest instance.

• You can use Classic Load Balancer (CLB) to access ASM gateway instances.

• The commercial edition of ASM Gateway supports TLS acceleration. It uses Intel's Multi-Buffer technology to accelerate HTTPS requests, improving QPS performance by 80% in tests.

Performance depends on manual tuning.

Supported mainstream protocols

Supports HTTP, HTTPS, QUIC, WebSocket, WSS, and gRPC protocols.

• Supports the HTTPS protocol and dynamic certificate loading.

• Lets you access internal gRPC services through the ingress gateway and switch traffic between two versions of a gRPC service.

• Supports protocol transcoding, which lets you use HTTP/JSON to access gRPC services within the service mesh.

• Lets you access WebSocket services within the mesh through the ingress gateway.

Supports HTTP, HTTPS, and gRPC protocols.

Observability

• Supports log collection through access logs and metrics.

• Supports integration with SLS for access logs.

• Supports integration with CloudMonitor for metrics.

• Supports alerting. You can connect the alerting feature to CloudMonitor.

• Provides an intuitive and easy-to-use visualization of the mesh topology for analysis.

• Supports integration with self-managed Prometheus for monitoring.

• Supports integration with ARMS.

• Supports integration with SLS.

• Supports custom monitoring metrics.

• Supports service-level objective (SLO) policies.

Collects logs through access logs.