Improve SSL-TPS performance on public cloud platforms
You can get better SSL-TPS performance on AWS and GCP clouds by distributing the packet engine (PE) weights equally. Enabling this feature might result in a slight drop in HTTP throughput by around 10–12 %.
On AWS and GCP clouds, the NetScaler VPX instances with 10–16 vCPUs do not show any performance improvement because the PE weights are equally distributed by default.
Note:
In the Azure cloud, the PE weights are equally distributed by default. This feature does not improve any performance for the Azure instances.
Configure PE mode by using the NetScaler CLI
After setting the PE mode, you must reboot the system for the configuration changes to take effect.
At the command prompt, type:
set cpuparam pemode [CPUBOUND | Default]
When the PE mode is set to CPUBOUND, the PE weights are equally distributed. When the PE mode is set to DEFAULT, the PE weights are set to default values.
Note:
This command is node specific. In a high availability or a cluster setup, you must run the command on each node. If you run the command on CLIP, the following error occurs:
Operation not permitted on CLIP
To show the state of the PE mode that is configured, run the following command:
show cpuparam
Example:
> show cpuparam
Pemode: CPUBOUND
Done
Apply PE mode configuration at the first boot of the NetScaler appliance in the cloud
To apply the PE mode configuration at the first boot of the NetScaler appliance in the cloud, you must create a
/nsconfig/.cpubound.conf file using the custom script. For more information, see Apply NetScaler VPX configurations at the first boot of the NetScaler appliance in cloud.