ELB Advantages

O&M

Fully managed and O&M-free

Manual installation, upgrade, and maintenance

Billing modes

• Elastic specifications: You are billed for how long each load balancer is running and the number of LCUs you use.
• Fixed specifications: Multiple specifications are available for you to select to best meet your needs. You are charged for the total LCUs you use.

You are billed for resources reserved for peak hours.

Deployment

• Deployed in clusters
• Multi-AZ

Deployed in VMs or containers

Reliability

• If there are traffic bursts, servers are added automatically.
• Node-level/AZ-level DR and 99.99% of SLA

• Sufficient computing resources need to be reserved to handle traffic surges during peak hours.
• Layer 7 performance depends on underlying computing resources. There is no SLA commitment.

Performance

ELB can handle up to tens of millions of concurrent connections and establish millions of new connections.

Only active/standby deployment is supported for Layer 4 load balancing. The performance is restricted by resource specifications.

Configuration change

Dynamic loading is supported.

• A reload process is required for configuration updates, which may interrupt persistent connections.
• A reload is required for changing Lua plug-ins.

SSL offloading

SSL encryption/decryption is performed load balancers, which relieves servers from decrypting or encrypting data.

SSL encryption/decryption is performed on backend servers, compromising server performance.

Related services

• Web Application Firewall (WAF) for protecting apps and websites against attacks
• Cloud Eye for monitoring cloud services and resources
• Log Tank Service (LTS) for collecting, querying, and storing access logs

Manual deployment is required for additional functions.

Each load balancer has exclusive access to isolated resources, allowing your services to handle a massive number of requests. A single load balancer deployed in an AZ can handle up to 20 million concurrent connections.

If multiple AZs are configured for a load balancer, its performance, such as the number of new connections and the number of concurrent connections, will be multiplied by the number of AZs. For example, if you configure two AZs for a dedicated load balancer, it can handle up to 40 million concurrent connections.

Dedicated load balancers can route traffic uninterruptedly. If servers in one AZ are unhealthy, they automatically route traffic to healthy servers in other AZs. Dedicated load balancers provide a comprehensive health check system to ensure that incoming traffic is only routed to healthy backend servers, which improves the availability of your applications.

Dedicated load balancers support TLS 1.3 and can route HTTPS requests to backend servers. You can select or customize security policies that fit your security requirements.

Dedicated load balancers support Quick UDP Internet Connection (QUIC), TCP, UDP, HTTP, and HTTPS, so that they can route requests to different types of applications.

Dedicated load balancers can route requests based on their content, such as the request method, header, URL, path, and source IP address. They can also redirect requests to another listener or URL, or return a fixed response to the clients.

Dedicated load balancers can route requests to both servers on the cloud and on premises, allowing you to leverage cloud resources to handle traffic bursts.

Dedicated load balancers provide a diverse set of algorithms that allow you to configure different traffic routing policies to meet your requirements while keeping deployments simple.

Shared load balancers provide guaranteed performance, which can handle up to 50,000 concurrent connections, 5,000 new connections per second, and 5,000 queries per second.

Shared load balancers can route traffic across AZs, ensuring that your services are uninterrupted. If servers in an AZ are unhealthy, ELB automatically routes traffic to healthy servers in other AZs. Shared load balancers provide a comprehensive health check system to ensure that incoming traffic is only routed to healthy backend servers, which improves the availability of your applications.

Shared load balancers support TCP, UDP, HTTP, and HTTPS protocols to route requests to different types of applications.

Shared load balancers provide a diverse set of algorithms that allow you to configure different traffic routing policies to meet your requirements while keeping deployments simple.

Load balancers can distribute across AZs more evenly.