I aim to implement a function as efficiently as possible: perform content matching on transmitted data, generate corresponding tag fields based on the different nature of the content (e.g., education-related, entertainment-related), and insert these tags into the reserved fields of the IP or transport layer.
In this scenario, can DOCA achieve full hardware support? Is PIPE only capable of matching simple fields like the five-tuple? Alternatively, is a hardware-software hybrid approach feasible—for instance, having software analyze the specific content and then hardware add the specific tags?
Hello,
DOCA offers a versatile framework for packet processing, enabling both hardware-accelerated and software-assisted pipelines.
Using DOCA Flow, applications can match various packet fields beyond the basic five-tuple and carry out actions such as header modification, metadata tagging, or making forwarding decisions.
For more advanced needs like content-based classification or tagging, a hybrid approach can be used, where software analyzes the payload and sets DOCA Flow rules to tag or modify headers for subsequent packets.
Please refer to the following documentation for more information:
• DOCA Overview
• DOCA Flow Programming Guide
We recommend making sure that your environment uses a supported software stack, including the operating system, kernel, driver, firmware, and DOCA version.
Thank you,
NVEX Networking Technical Support Team