Pipeline instances

By default, the DNCA is configured to create a single instance of the DNCA pipelined process for use by all instances of the DNCA application. If needed, more instances of the pipelined process can be created to more efficiently distribute the work across available resources.

Note: Multiple instances of the DNCA pipeline are available in DNCA Build.

When multiple instances of the pipelined process are created, each instance of the DNCA application submits the TCP packets. These packets are assembled by its reassd process to the pipeline queue. This queue then distributes the packets for processing to the configured pipelined instances in a round-robin fashion.

The effectiveness of multiple pipelines depends on the number of available CPU cores. The general rule of thumb is to allocate one available CPU core for each pipeline instance. If four idle CPU cores are available, you must configure no more than four pipeline instances if you expect a full CPU load.

Note: You must always leave at least one CPU core in the machine for system-based processes.
  • The theoretical maximum limit on the number of pipelines is 128. It is not likely you must approach this limit.
Note: All pipelines operate on the same set of configuration rules. There must be no differences between privacy rules or other configuration options between multiple pipelines.
Setting
Description
Instances
Number of instances of the pipelined process.
  • To create a different number of instances, enter a value in the textbox and click Save Changes.
    Note: Be conservative when creating new instances of the pipelined. Increment instances by one and then assess the results and impact on available resources on the DNCA server.

To retrieve statistics on an individual pipeline instance, run the following commands. Instance #0 stats:


dcxstats instdata

Instance #N stats


dcxstats -I<#N> instdata

where

  • <#N> is the instance identifier.