Configuring Server Load Balancing Server Load Balancing Overview
OmniSwitch 6800/6850/9000 Network Configuration Guide March 2008 page 34-7
Server Load Balancing Overview
In the current release, you can configure up to 16 Server Load Balancing (SLB) clusters on an
OmniSwitch 6850/9000. Each cluster may consist of 16 servers, which allows for configuration of up to
256 physical servers per switch. The following sections describe SLB operational theory (see “Server
Load Balancing Cluster Identification” on page 34-7), an SLB example (“Server Load Balancing Exam-
ple” on page 34-8), and server health monitoring (see “Server Health Monitoring” on page 34-9).
Note. Alcatel-Lucent also offers link aggregation, which combines multiple Ethernet links into one virtual
channel. Please refer to Chapter 17, “Configuring Dynamic Link Aggregation,”for more information on
link aggregation and dynamic link aggregation, and to Chapter 16, “Configuring Static Link Aggrega-
tion,” for information on static (OmniChannel) link aggregation.
Server Load Balancing Cluster Identification
An SLB cluster consists of a group of physical servers, also known as a server farm. The SLB cluster
appears as one large virtual server, which is identified using one of the following methods:
• Virtual IP (VIP)—An IP address is assigned to the cluster (virtual server). Client requests destined for
this VIP are routed (Layer-3 mode) to the servers that are members of the VIP cluster. Note that it is
necessary to configure cluster servers with a loopback interface.
• Condition—A QoS policy condition name is assigned to the cluster (virtual server). Client requests that
meet the criteria of the policy condition are bridged (Layer-2 mode) or routed (Layer-3 mode) to the
servers that are members of the condition cluster. Note that it is not necessary to configure cluster serv-
ers with a loopback interface.
Note. See “Configuring the Server Farm” on page 34-10 for more information on configuring servers. See
“Configuring and Deleting SLB Clusters” on page 34-24 for more information on configuring VIP and
condition clusters.
Server Load Balancing Cluster Modes
The cluster mode refers to whether client requests are bridged (Layer-2 mode) or routed (Layer-3 mode)
by the switch to the appropriate SLB cluster. A VIP cluster only supports Layer-3 mode, so request pack-
ets are always routed to the cluster. A condition cluster supports both Layer-2 and Layer-3 modes.
When the Layer-3 mode is active (VIP or condition clusters), routed packets are modified as follows:
• The source MAC address is changed to the MAC address for the switch router interface.
• The destination MAC address is changed to the MAC address of the destined server.
• The TTL value is decremented.
When the Layer-2 mode is active (condition clusters only), request packets are not modified and are only
switched within the same VLAN domain. The Layer-2 or Layer-3 mode is selected when the condition
cluster is configured on the switch. See “Configuring an SLB Cluster with a QoS Policy Condition” on
page 34-24 for more information.
To Next Page
Loading...
