Command differences in Load Balancer IPv4 and IPv6

Cluster add command

• The load balancer for IPv4 and IPv6 does not have a limitation on maximum ports and server in a cluster, hence maxserver and maxport options are not available.
• The staletimeout and stickytimeout commands are available only at port level configuration.
• The primaryhost option is not valid since mutual high availability is not supported in load balancer for IPv4 and IPv6.
• The porttype command is not valid since Load Balancer for IPv4 and IPv6 supports only TCP applications.

Executor commands

• The staletimeout and stickytimeout commands are available only at port level configuration.
• The Porttype command is not valid since Load Balancer for IPv4 and IPv6 supports only TCP applications.
• NFA autodetection and manual configuration option is available in Load Balancer for IPv4 and IPv6. In Load Balancer for IPv4, manual configuration is the only option.
• No limit on maximum clusters, port and servers, hence these options are not available in Load Balancer for IPv4 and IPv6.
• The hasynctimeout command is not supported. When you run Load Balancer for IPv4 with high availability and partners synchronize, this parameter controls how long that sync is allowed to take. Many packets flow and it can consume a long time. If the sync does not finish, high availability does not work. Hence, this option is discontinued.
• The mss command is not available in Load Balancer for IPv4 and IPv6 because it is handled by ICMP-based fragmentation.
• The Sharedbandwidth command was added to give customers who charge clients for service the ability to deliver only bandwidth that is paid for. This method is problematic because the load balancer ensures that only the bandwidth is not exceeded and not that it is available for the client since bandwidth shared with all other traffic on subnet. Use data/tooling at the backend servers to charge for service and not try to control at the load balancer. This option is not implemented in Load Balancer for IPv4 and IPv6.
• The wideport number is not supported. This command is deprecated in Load Balancer for IPv4. This option was used to allow load balancer to encapsulate and forward to another load balancer on a remote subnet. The addition of NAT forwarding mechanism rendered this option redundant and it is no longer available.
• The address command is not available in Load Balancer for IPv4 and IPv6.
• Mutual high availability is not supported, hence the both option is not available in dscontrol high availability backup add command.

Port level commands

• The forwarding method is at the server level, therefore the method option is not available.
• There are no limitations on number of servers in each port, and therefore, maxservers is not available.
• Load Balancer for IPv4 and IPv6 supports only TCP applications, therefore, port type is not valid.
• Denial of service is not supported and therefore, maxhalfopen and halfopenaddress report options are not available.
• The protocol option is used for CBR forwarding and therefore available as part of cbrcontrol command.
• Cross-port affinity cannot be modified with the set command in the Load Balancer for IPv4 and IPv6. If cross-port affinity is defined on the port, the port must be removed and readded to change the crossport value.
• In Load Balancer for IPv4, because the affinity selection algorithm was not available, during replication, both connection and affinity records were always sent. In load balancer for IPv4 and IPv6, there are different replication strategies, such as affinity and connection plus affinity algorithms. Hence, the values for repstrategy in Load Balancer for IPv4 and IPv6 are also different.

High availability replication is disabled by default in the Load Balancer for IPv4 and IPv6.

Rule commands

Load Balancer for IPv4 and IPv6 provides only active and true rules. It does not provide the following parameters:

• IP
• Time
• Connection
• Port
• Service
• Reserved bandwidth, share level, and shared bandwidth
• Content

Server commands

• The up and down commands are not supported in Load Balancer for IPv4 and IPv6.
• Configure forwarding method at the server level in Load Balancer for IPv4 and IPv6. The option to specify forwarding method is available when you add a server. The default value is MAC forwarding. You need to configure encapsulation and NAT forwarding.
• The router address that you configure for NAT is not a settable parameter.
• The mapport option is not available with the dispatcher component because NAPT is not implemented in Load Balancer for IPv4 and IPv6.

Help commands

dscontrol manager

The quiesce feature differs between the two load balancers. Load Balancer for IPv4 has only the quiesce now option. If you do not specify now, existing connections complete and forward subsequent new connections to the quiesced server from those clients with existing connections that are designated as sticky, if the quiesced server receives the new request before stickytime expires. However, if you have not enabled the sticky (affinity) feature, the quiesced server cannot receive any new connections. By specifying now, you quiesce the server so existing connections can complete, but forbids all new connections, including subsequent new connections, from those clients with existing connections that are designated as sticky.

Load Balancer for IPv4 and IPv6 has two options: quiesce server and Cluster@Port@Server. The former quiesces the server on all ports and all clusters and the latter quiesces the server on a specific cluster and port combination. The existing connections are served and new connections are not allowed on the quiesced server, therefore, the behavior is similar to the quiesce now option in Load Balancer for IPv4. There is no option to change this quiesce behavior.

Difference in load balancing algorithms

• Load Balancer for IPv4 provides affinity based on a sticky time, which starts after a connection terminates. If a client requests a new connection during the sticky time, the request is forwarded to the previously selected server. The Load Balancer for IPv4 and IPv6 maintains affinity based upon the connection selection algorithm and the sticky time. When the selection algorithm is set to affinity or conn+affin, any client request received within the sticky time period is sent to the same server. The sticky time begins when a client connection starts rather than when the connection terminates.
• There is a difference in the way the round robin load balancing algorithm is implemented.

  For example, if server A has a server weight of 5 and server B has a server weight 10, in Load Balancer for IPv4 and IPv6, the first five requests might be forwarded to server A and then the next 10 requests are forwarded to server B, or the converse. The second server is chosen when the weight of the previous server is fully exhausted, and the cycle repeats.

  For Load Balancer for IPv4, the requests are forwarded alternately among server A and B. Considering a weight of 5 for server A and 10 for server B, the first 10 requests are forwarded alternately between server A and B (server A and server B serve five requests each). Because the weight of server A is reached, the next five requests are served only by server B. Now that the weight of server B is also exhausted, the cycle repeats.

Scripts