DPDK patches and discussions
 help / color / mirror / Atom feed
From: "De Lara Guarch, Pablo" <pablo.de.lara.guarch@intel.com>
To: Nishant Verma <vnish11@gmail.com>
Cc: "dev@dpdk.org" <dev@dpdk.org>
Subject: Re: [dpdk-dev] Fwd: port-queue-process
Date: Sun, 29 Nov 2015 20:43:59 +0000	[thread overview]
Message-ID: <E115CCD9D858EF4F90C690B0DCB4D8973C861BED@IRSMSX108.ger.corp.intel.com> (raw)
In-Reply-To: <CAHhCjUFHrKtF_6+VqYhpgUDKM1b_q5RWamp-ADqDLFH9ung+fg@mail.gmail.com>

Hi,



> From: Nishant Verma [mailto:vnish11@gmail.com]

> Sent: Saturday, November 28, 2015 7:28 PM

> To: De Lara Guarch, Pablo

> Cc: dev@dpdk.org<mailto:dev@dpdk.org>

> Subject: Re: [dpdk-dev] Fwd: port-queue-process

>

> Thanks for the reply.

> Got your point. It means, suppose DPDK app is getting packet from 10

> different source. On very first packet from any source, we first check source

> MAC address and then search what queue we have used last time, and

> placed in the same else choose the queue based on some factors.?



What do you mean with sources? Are you getting all packets in the same port, right?

If so, you can identify the sources by the differences between packets, such as the L2, L3… headers.

It looks like you want to add rules dynamically, so at first you don’t know what your sources are,

but as you get packets from different sources, you can assign them to different cores.

In that case, yes, you could see what queues you used so far, and assign it to a queue that is not being used,

but that depends on what the core is going to do with that flow (from a source).

I mean, if you are going to run the same code for all packets, but in different cores,

then probably something like RSS can work for you, but if the code is going to be different

depending on the source, then you need to know in advance these sources

and program flow director accordingly.



> One more thing, suppose if DPDK app is getting packet from one single

> source and that source is handling 10 different source? In that case how we

> make diffenence?



In this case, I guess your worry is that you would have same L2 header for all sources?

As long as you can identify your source in any field of the packet (probably L3 header),

it doesn’t matter if you are in the first scenario or this one.



Pablo



> Thanks.

>

> On Sat, Nov 28, 2015 at 2:12 PM, De Lara Guarch, Pablo

> <pablo.de.lara.guarch@intel.com<mailto:pablo.de.lara.guarch@intel.com>> wrote:

> Hi Nishant,

>

> > -----Original Message-----

> > From: dev [mailto:dev-bounces@dpdk.org] On Behalf Of Nishant Verma

> > Sent: Saturday, November 28, 2015 5:42 AM

> > To: dev@dpdk.org<mailto:dev@dpdk.org>

> > Subject: [dpdk-dev] Fwd: port-queue-process

> >

> > Hi All,

> >

> > I have one doubt about dpdk.

> >

> > Question. Suppose on 8 core machine DPDK process is running on all

> cores,

> > all process has its own Rx & Tx queue. My question is, let we have a

> packet

> > from some network for DPDK application running on 2nd core, how it will

> > reach to 2nd core and not to some other core.

> >

> > Any clue and reference will be appreciated. It might seems to little doubt

> > but as i don't know, its big for me.

>

> I think what you are looking for is packet filtering and queue assignment,

> specifically, Flow Director. I assume you have one port and several queues

> in that port:

> you will need assign each queue to a different core, and program the device

> in a way that

> packets with a particular source MAC address or IP address will go to a

> specified queue,

> so the desired core can poll the packets from that queue.

>

> Let me know if you have any more questions,

>

> Pablo

>

> >

> > ​thanks.

> >

> > --

> > Rgds,

> >

> > NV​

>

>

>

> --

> Rgds,

> ​NV




      reply	other threads:[~2015-11-29 20:44 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
     [not found] <CAHhCjUF9P_gC4eXio8US0=RXcvkhA=_now10JrioP-ft20fK0g@mail.gmail.com>
2015-11-28  5:41 ` Nishant Verma
2015-11-28 19:12   ` De Lara Guarch, Pablo
2015-11-28 19:27     ` Nishant Verma
2015-11-29 20:43       ` De Lara Guarch, Pablo [this message]

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=E115CCD9D858EF4F90C690B0DCB4D8973C861BED@IRSMSX108.ger.corp.intel.com \
    --to=pablo.de.lara.guarch@intel.com \
    --cc=dev@dpdk.org \
    --cc=vnish11@gmail.com \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).