Product Documentation
Download:PDFPRINT

Client Keep-Alive

Aug 06, 2014

The client keep-alive feature enables multiple client requests to be sent on a single client connection. This feature helps in a transaction management environment where typically the server closes the client connection after serving the response. The client then opens a new connection for each request and spends more time on the transaction.

Client keep-alive resolves this issue by keeping the connection between the client and the appliance (client-side connection) open even after the server closes the connection with the appliance. This allows sending multiple client requests using a single connection and saves the round trips associated with opening and closing a connection. Client keep-alive is most beneficial in SSL sessions.

Client keep-alive is also useful under either of the following conditions:
  • When the server does not support client keep-alive.
  • When the server supports client keep-alive but an application on the server does not support client keep-alive.
Note: Client keep-alive is applicable for HTTP and SSL traffic.

Client-keep alive can be configured globally to be able to handle all traffic. It can also be configured to be active only on specific services.

In client keep-alive environment, the configured services intercept the client traffic and the client request is directed to the origin server. The server sends the response and closes the connection between the server and the appliance. If a "Connection: Close" header is present in the server response, the appliance corrupts this header in the client-side response, and the client-side connection is kept open. As a result, the client does not have to open a new connection for the next request; instead, the connection to the server is reopened.

Note: If a server sends back two "Connection: Close" headers, only one is edited. This results in significant delays on the client rendering of the object because a client does not assume that the object has been delivered completely until the connection is actually closed.

Configuring Client Keep-Alive

Updated: 2014-08-12

Client keep-alive, by default, is disabled on the NetScaler, both globally and at service level. Therefore, you must enable the feature at the required scope.

Note: If you enable client keep-alive globally, it is enabled for all services, regardless of whether you enable it at the service level.

Additionally, if required, you can configure some HTTP parameters to specify the maximum number of HTTP connections retained in the connection reuse pool, enable connection multiplexing, and enable persistence Etag.

Note: When Persistent ETag is enabled, the ETag header includes information about the server that served the content. This ensures that cache validation conditional requests or browser requests, for that content, always reaches the same server.

To configure client keep-alive by using the command line interface

At the command prompt, do the following:

  1. Enable client keep-alive on the NetScaler.
    • At global level

      enable ns mode cka

    • At service level

      set service <name> -CKA YES

      Note: Client keep-alive can be enabled only for HTTP and SSL services.
  2. Configure the required HTTP parameters on the HTTP profile that is bound to the service(s).

    set ns httpProfile <name> -maxReusePool <value> -conMultiplex ENABLED -persistentETag ENABLED

    Note: Configure these parameters on the nshttp_default_profile HTTP profile, to make them available globally.

To configure client keep-alive by using the configuration utility

  1. Enable client keep-alive on the NetScaler.
    • At global level

      Navigate to System > Settings, click Configure Modes and select Client side Keep Alive.

    • At service level

      Navigate to Traffic Management > Load Balancing > Services, and select the required service. In the Settings grouping, enable Client Keep-Alive.

  2. Configure the required HTTP parameters on the HTTP profile that is bound to the service(s).

    Navigate to System > Profiles, and on HTTP Profiles tab, select the required profile and update the required HTTP parameters.