FW: Our UPF Features

Skip to first unread message


Jan 1, 2024, 10:11:04 AMJan 1
to p4-dev, p4-design

Any operator can get up and running with our upf6 using x86 or ARM machine. Telcos prefer vpp, both Cisco and Mavenir sell upf using vpp.  My company will give our upf6 free to test and deploy to an operator. The product is generated by our compiler. We can discuss support.






From: hemant via P4 VPP <p4-...@googlegroups.com>
Sent: Monday, January 01, 2024 8:59 AM
To: 'P4 VPP' <p4-...@googlegroups.com>
Subject: Our UPF Features


Our UPF Features

UPF is supported in following modes:

(a) Branching Point (BP) for multi-homing.

(b) Anchor for multi-homing.

(c) Intermediate node with a teid and qfi filter table.

The UPF supports N3, N9, and N6 interfaces for user plane and N4 interface to SMF.

General features supported by our p4c:

(a) punt path, drop, counters per table, node trace, and node errors via node trace.

(b) CLI and API are generated for any plugin. For UPF plugin, manual Python script was written to test UPF and also test all API.

PDR (IP Filter Set is a 6-tuple for ipv4/ipv6), FAR, URR, Qos, ppi, gtpu tunnel encap/decap, gtp extensions, and replying to gtp echo request are supported for both IPv4 and IPv6. Each feature has PTF test with packets to show the feature is working in the data plane. For URR (Usage reports), the plan is to have the control-plane collect data plane counters for each table entry periodically.


You received this message because you are subscribed to the Google Groups "P4 VPP" group.
To unsubscribe from this group and stop receiving emails from it, send an email to p4-vpp+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/p4-vpp/1e8301da3cba%249c0c3400%24d4249c00%24%40mnkcg.com.
For more options, visit https://groups.google.com/d/optout.

Reply all
Reply to author
0 new messages