Product Introduction
Updated at：2025-10-16
Load Balancer Instance
Baidu Load Balance (BLB) consolidates multiple Baidu Cloud Computes within the same region into a group, assigns an intranet or Internet service address, and distributes concurrent front-end requests across multiple real cloud servers. This balances application traffic, facilitates horizontal scaling of service performance, and enhances availability by automatically resolving single points of failure.
It now launches application BLB instance, and its features the following deployment architecture in a specific scenario:
Instance type comparison:
| Instance type | Description | Differences |
|---|---|---|
| General-purpose BLB | The service includes standard load balancer components, such as listeners and real servers, to address general load balancing needs. It supports the distribution of IPv4 or IPv6 traffic from the Internet or intranet to multiple IPv4 real servers using specific load balancing protocols, offering a lightweight, straightforward, and cost-effective solution. | Real servers are organized by instance dimension. Each listener can forward to all real servers of the instance, with identical real ports for corresponding listeners. Traffic forwarding methods: Unified listener scheduling supports weighted round-robin, weighted least connections, and traffic forwarding based on source IP. |
| Application BLB | The service is designed for complex application deployments, particularly large-scale websites. By employing a policy-based traffic forwarding framework, it can distribute IPv4 or IPv6 traffic from the Internet or intranet to various IPv4 real server groups based on predefined strategies, achieving service-oriented traffic load balancing. | It supports server groups, allows the organization of real servers by listener dimension. Each listener can correspond to different real server groups via policies. The IP group scenario supports setting different real ports for servers within the same server group. Traffic forwarding method: Policy-based forwarding by domain name/URL/source IP/Cookie; policy-based weight setting also supported. |
| General-purpose IPv6 BLB | It incorporates standard load balancer features, such as listeners and real servers, to address users' general load balancing demands. The service supports the distribution of IPv6 traffic from the Internet or intranet to multiple IPv6 real servers using specific load balancing protocols, ensuring a lightweight design and straightforward setup. | Real servers are organized by instance dimension. Each listener can forward to all real servers of the instance, with identical real ports for corresponding listeners. Traffic forwarding methods: Unified listener scheduling supports weighted round-robin, weighted least connections, and traffic forwarding based on source IP. Only support distributing front-end IPv6 traffic to IPv6 real servers. |
| Application IPv6 BLB | Tailored for complex application deployments, particularly for large-scale websites, this service uses a policy-driven traffic forwarding framework to route IPv6 traffic from the Internet or intranet to multiple IPv6 real server groups based on relevant strategies, enabling efficient service-based traffic load balancing. | It supports server groups, allows the organization of real servers by listener dimension. Each listener can correspond to different real server groups via policies. The IP group scenario supports setting different real ports for servers within the same server group. Traffic forwarding method: Policy-based forwarding by domain name/URL/source IP/Cookie; policy-based weight setting also supported. Only support distributing front-end IPv6 traffic to IPv6 real servers. |
Specification comparison:
| Instance specification. | Performance limit |
|---|---|
| Standard 1 | Concurrent connections: 50,000; new connections per second: 5,000; queries-per-second: 5,000; bandwidth limit: 1Gbps |
| Standard 2 | Concurrent connections: 100,000; new connections per second: 10,000; queries-per-second: 10,000; bandwidth limit: 2Gbps |
| Enhance 1 | Concurrent connections: 200,000; new connections per second: 20,000; queries-per-second: 20,000; bandwidth limit: 4Gbps |
| Enhance 2 | Concurrent connections: 500,000; new connections per second: 50,000; queries-per-second: 30,000; bandwidth limit: 6Gbps |
| Ultra Large 1 | Concurrent connections: 1,000,000; new connections per second: 100,000; queries-per-second: 50,000; bandwidth limit: 10Gbps |
| Ultra Large 2 | Concurrent connections: 2,000,000; new connections per second: 200,000; queries-per-second: 100,000; bandwidth limit: 20Gbps |
| Ultra Large 3 | Concurrent connections: 4,000,000; new connections per second: 400,000; queries-per-second: 200,000; bandwidth limit: 40Gbps |
| Ultra Large 4 | Concurrent connections: 6,000,000; new connections per second: 600,000; queries-per-second: 300,000; bandwidth limit: 60Gbps |
| Ultra Large 5 | Concurrent connections: 8,000,000; new connections per second: 800,000; queries-per-second: 800,000; bandwidth limit: 80Gbps |
Load Balance Dedicated Cluster
Load Balance Dedicated Cluster facilitates the deployment of load balancer instances within dedicated clusters, ensuring controlled performance, exclusive resource allocation, and physical isolation. This service caters to users requiring ultra-high performance and exclusive resources.
Notes If a user doesn't purchase a Load Balance Dedicated Cluster, BLB instances he/she created are located in a shared cluster by default, where instance performance is shared by multiple tenants. Performance may be constrained during resource contention. If the user requires ultra-high performance and exclusive resources, we recommended that you use a Load Balance Dedicated Cluster.
Difference between load balance dedicated clusters and shared clusters:
| Comparison items | Description | Dedicated cluster | Shared cluster |
|---|---|---|---|
| Dedicated configuration | Dedicated special service requirement configuration | Support | Not supported |
| Traffic isolation | Traffic isolation from other users' services | Support | Not supported |
| Precise traffic routing | Precisely routes service traffic to this cluster | Support | Not supported |
| Service isolation | Deployment isolation from other users' services | Support | Not supported |
| Cluster performance | Overall cluster performance limit | Support specifying performance specifications and one-click scale-up and dilatation | A limit is set for the cluster performance; multiple users compete for resources when encountering performance bottlenecks. |
| Cluster monitoring | Monitor service load at the cluster granularity | Support | Not supported |
| Associated instances | Support viewing all instances on the cluster | Support | Not supported |
| Price | Charged by cluster in prepay mode; BLB instances will no longer be charged. | BLB configuration fee is charged only |