Does your Java program need to see those protocol fields from every packet, or would a 1:N random sample be enough for the analysis? (If so, see sFlow plugin).
Neil On Tue, Jul 28, 2026 at 11:12 PM Benoit Ganne (bganne) via lists.fd.io <[email protected]> wrote: > The typical high-performance construct for this kind of usecase should be > a read-only ring-buffer in shared memory where VPP can expose the extracted > metadata and the Java process can read them. > That would avoid additional memory copies while still keeping memory > isolation. > I suspect the eqsiest is probably to frame your own protocol on top of > memif, look at libmemif in the VPP repository. > > Best > ben > > ________________________________________ > From: [email protected] <[email protected]> on behalf of > [email protected] <[email protected]> > Sent: Tuesday, July 28, 2026 10:59 > To: vpp-dev > Subject: [vpp-dev] Efficiently exporting extracted protocol fields from > VPP data plane to a local Java application – seeking advice > > > Hello VPP community, > > We are currently working with VPP versions 22.02 and 26.06, and we have a > requirement to analyze data traffic on a 10G NIC interface in the VPP data > plane. Specifically, we need to extract certain protocol fields from the > packets and then efficiently deliver the extracted data to a Java > application running on the same machine. > > We would greatly appreciate any suggestions, best practices, or existing > plugin examples that address similar use cases. A > > Thank you in advance for your insights! > > > > > > >
-=-=-=-=-=-=-=-=-=-=-=- Links: You receive all messages sent to this group. View/Reply Online (#27134): https://lists.fd.io/g/vpp-dev/message/27134 Mute This Topic: https://lists.fd.io/mt/120481202/21656 Group Owner: [email protected] Unsubscribe: https://lists.fd.io/g/vpp-dev/leave/14379924/21656/631435203/xyzzy [[email protected]] -=-=-=-=-=-=-=-=-=-=-=-
