Alibaba Cloud Load Balancing SLB is divided into traditional load balancing CLB (former SLB), application load balancing ALB and network load balancing NLB. What are the differences between the three? CLB is the previous traditional SLB, a 4-layer load balancing based on physical machine architecture; ALB is an application load balancing, a 7-layer load balancing; NLB is a network load balancing, a 4-layer load balancing based on the NFV virtualization platform. Alibaba Cloud Encyclopedia will explain in detail the differences between Alibaba Cloud load balancing SLB, ALB and NLB and the selection method:
Alibaba Cloud Load Balancing SLB
Alibaba Cloud load balancing SLB (Server Load Balancer) is divided into three types: network load balancing NLB, application load balancing ALB and traditional load balancing CLB (original SLB). Initially Alibaba Cloud only had one traditional load balancing SLB, this one The SLB instance is a four-layer load balancing; later, a seven-layer load balancing ALB was launched based on the traditional SLB, and the original four-layer SLB was renamed CLB; and later a network-based load balancing NLB based on the NFV virtualization platform that supports elastic scaling was launched. Many users don’t know what the difference is and how to choose? Alibaba Cloud Encyclopedia shares a comparison table of CLB, ALB and NLB:
Reference: https://dashi.aliyun.com/site/cloud/slb
| Load balancing comparison | Network load balancing NLB | Application load balancing ALB | Traditional load balancing CLB (formerly SLB) |
|---|---|---|---|
| position | Super layer 4 processing capabilities and large-scale SSL offloading capabilities focus on TCP, UDP and TCPSSL protocols for network layer delivery | Powerful 7-layer load balancing processing capabilities and rich advanced routing functions focus on HTTP, HTTPS and QUIC application layer protocols for application layer delivery | Powerful 4-layer load balancing processing capability supports TCP/UDP/HTTP/HTTPS and other protocols and is mainly delivered to the network layer |
| performance | Based on the NFV virtualization platform, it supports elastic scaling and a single instance can support up to 100 million concurrent connections. | Based on the NFV virtualization platform, it supports elastic scaling and a single instance can support up to 1 million QPS. | Based on physical machine architecture, a single instance supports up to 1 million concurrencies and 50,000 QPS. |
| Availability Commitment SLA | 99.995% (SLA provided after public beta) | 99.995％ | 99.95% |
| Operation and maintenance method | Network-based load balancing processing capabilities automatically and elastically scale with business scale without the need for peak and specification predictions. | Application load balancing processing capabilities automatically and elastically scale with business scale without the need to estimate peak values and specifications. | Traditional load balancing is sold according to specifications, and the peak bandwidth needs to be estimated based on the business scale. |
| Cloud native support | Coming soon | The official cloud-native Ingress gateway supports traffic splitting, mirroring, grayscale publishing and load balancing capabilities for blue-green testing. | Traditional load balancing supports cloud-native business scenarios and can be used in conjunction with container services such as Alibaba Cloud ACK/ASK. |
| Typical application scenarios | Scenario 1: Four-layer high-traffic and high-concurrency business scenario Scenario 2: Internet of Things, Internet of Vehicles and other IoT business portals Scenario 3: Multi-active disaster recovery, IDC cloud portal scenario | Scenario 1: Internet application 7-layer high-performance automatic elastic load balancing scenario Scenario 2: Audio and video application large traffic low-latency load balancing scenario Scenario 3: Cloud native application Canary Blue and Green release load balancing scenario | Scenario 1: Website/system high-reliability load balancing scenario in the same region Scenario 2: Layer 4 traffic heavy concurrent business load balancing scenario Scenario 3: In-city active-active/cross-region disaster recovery load balancing scenario |
In summary, Alibaba Cloud Load Balancing CLB is a traditional 4-layer load balancing instance. CLB is based on a physical machine architecture, and a single instance supports up to 1 million concurrencies and 50,000 QPS. Application load balancing ALB is a 7-layer load balancing, based on NFV virtualization. The platform supports elastic scaling, with a single instance supporting a maximum of 1 million QPS; the network load balancing NLB is a 4-layer load balancing, based on the NFV virtualization platform, supporting elastic scaling, and a single instance supports a maximum of 100 million concurrent connections.
how to choose? For four-layer load balancing in the same city, active-active, same-region load or cross-region disaster recovery load, choose traditional load balancing CLB (former SLB); for seven-layer load balancing audio and video applications, choose application load balancing ALB; for four-layer large traffic and high concurrency, NLB is the first choice for IoT business portals such as the Internet of Things and the Internet of Vehicles.
For more information about Alibaba Cloud Load Balancing SLB application scenarios and accurate quotations, please refer to the official page: Alibaba Cloud Load Balancing SLB Official Detailed Explanation