F5 Load Balancing Methods / Algorithms

Using the default load balancing method

The default load balancing method for the LTM system is Round Robin, which simply passes each new connection request to the next server in line. All other load balancing methods take server capacity and/or status into consideration.

If the equipment that you are load balancing is roughly equal in processing speed and memory, Round Robin mode works well in most configurations. If you want to use the Round Robin method, you can skip the remainder of this section, and begin configuring other pool settings that you want to add to the basic pool configuration.

Selecting a load balancing method

If you are working with servers that differ significantly in processing speed and memory, you may want to switch to one of the Ratio or dynamic methods.

Round Robin

This is the default load balancing method. Round Robin mode passes each new connection request to the next server in line, eventually distributing connections evenly across the array of machines being load balanced. Round Robin mode works well in most configurations, especially if the equipment that you are load balancing is roughly equal in processing speed and memory.

Ratio (member) and Ratio (node)

The LTM system distributes connections among machines according to ratio weights that you define, where the number of connections that each machine receives over time is proportionate to a ratio weight you define for each machine. These are static load balancing methods, basing distribution on static user-assigned ratio weights that are proportional to the capacity of the servers. 

Regarding Ratio load balancing:

Load balancing calculations may be localized to each pool (member-based calculation) or they may apply to all pools of which a server is a member (node-based calculation). This distinction is especially important with the Ratio method; with the Ratio (member) method, the actual ratio weight is a member setting in the pool definition, whereas with the Ratio (node) method, the ratio weight is a setting of the node.

The default ratio setting for any node is 1. If you use the Ratio (as opposed to Ratio (member) load balancing method, you must set a ratio other than 1 for at least one node in the configuration. If you do not change at least one ratio setting, the load balancing method has the same effect as the Round Robin load balancing method.

Warning: If you set the load balancing method to Ratio (node), as opposed to Ratio (Member), you must define a ratio setting for each node.

Dynamic Ratio

The Dynamic Ratio method is like the Ratio method except that ratio weights are based on continuous monitoring of the servers and are therefore continually changing.

This is a dynamic load balancing method, distributing connections based on various aspects of real-time server performance analysis, such as the current number of connections per node or the fastest node response time.

The Dynamic Ratio method is used specifically for load balancing traffic to RealNetworks® RealSystem® Server platforms, Windows® platforms equipped with Windows Management Instrumentation (WMI), or any server equipped with an SNMP agent such as the UC Davis SNMP agent or Windows 2000 Server SNMP agent. To implement Dynamic Ratio load balancing, you must first install and configure the necessary server software for these systems, and then install the appropriate performance monitor. For more information, see Appendix A, Additional Monitor Considerations .

Fastest (node) and Fastest (application)

The Fastest methods pass a new connection based on the fastest response of all currently active nodes. These methods may be particularly useful in environments where nodes are distributed across different logical networks. Load balancing calculations may be localized to each pool (member-based calculation) or they may apply to all pools of which a server is a member (node-based calculation).

Least Connections (member) and Least Connections (node)

The Least Connections methods are relatively simple in that the LTM system passes a new connection to the node that has the least number of current connections. Least Connections methods work best in environments where the servers or other equipment you are load balancing have similar capabilities.

These are dynamic load balancing methods, distributing connections based on various aspects of real-time server performance analysis, such as the current number of connections per node or the fastest node response time.

Load balancing calculations may be localized to each pool (member-based calculation) or they may apply to all pools of which a server is a member (node-based calculation).

Observed (member) and Observed (node)

The Observed methods use a combination of the logic used in the Least Connections and Fastest modes. With the Observed methods, nodes are ranked based on a combination of the number of current connections and the response time. Nodes that have a better balance of fewest connections and fastest response time receive a greater proportion of the connections. The Observed modes also work well in any environment, but may be particularly useful in environments where node performance varies significantly.

These are dynamic load balancing methods, distributing connections based on various aspects of real-time server performance analysis, such as the current number of connections per node or the fastest node response time.

Load balancing calculations may be localized to each pool (member-based calculation) or they may apply to all pools of which a server is a member (node-based calculation).

Predictive (member) and Predictive (node)

The Predictive methods also use the ranking methods used by the Observed methods, where nodes are rated according to a combination of the number of current connections and the response time. However, with the Predictive methods, the LTM system analyzes the trend of the ranking over time, determining whether a node's performance is currently improving or declining. The nodes with better performance rankings that are currently improving, rather than declining, receive a higher proportion of the connections. The Predictive methods work well in any environment.

The Predictive methods are dynamic load balancing methods, distributing connections based on various aspects of real-time server performance analysis, such as the current number of connections per node or the fastest node response time.

Load balancing calculations may be localized to each pool (member-based calculation) or they may apply to all pools of which a server is a member (node-based calculation).