DPDK patches and discussions
 help / color / mirror / Atom feed
From: Stephen Hemminger <stephen@networkplumber.org>
To: satish <nsatishbabu@gmail.com>
Cc: dev@dpdk.org
Subject: Re: [dpdk-dev] Dynamic port/pipe QoS configuration
Date: Sat, 18 Oct 2014 11:16:43 +0200	[thread overview]
Message-ID: <20141018111643.44a9d59f@uryu.home.lan> (raw)
In-Reply-To: <CADVv77pEx1-ihzFHcj1bXu_7YBSzcvwUjpBv73pGLU+raDM+iA@mail.gmail.com>

On Fri, 17 Oct 2014 14:59:35 -0700
satish <nsatishbabu@gmail.com> wrote:

> Hi,
> Can someone please respond to below?
> Thank you.
> 
> On Mon, Oct 13, 2014 at 3:54 PM, satish <nsatishbabu@gmail.com> wrote:
> 
> > Hi,
> > We are trying to provide QoS support for one of our clients using
> > rte_sched.
> > In our implementation we are treating each pipe as a customer.
> > So, we can have maximum of 4096 customers per sub-port.
> > Customers(pipe) can be added, deleted or modified dynamically. Each
> > customer can have different profiles.
> >
> > Currently we are using DPDK-v1.6.
> > Can I modify pipe profile during run time using rte_sched_pipe_config ()?
> >
> > Our plan is to have initial configs as below (similar to examples in DPDK)
> > [1] Specify port params at the initialization of port as below
> > static struct rte_sched_port_params port_param = {
> > :
> > :
> > .n_subports_per_port = 1,
> > .n_pipes_per_subport = 4096,
> > .qsize = {64, 64, 64, 64},
> > .pipe_profiles = pipe_profile,
> > .n_pipe_profiles = 1,
> > }
> >
> > [2]
> > static struct rte_sched_subport_params subport_param[] = {
> > {
> > .tb_rate = Link speed (1G/10G..) divided by 8 (bits),
> > .tb_size = 1000000,
> >
> > .tc_rate = {Same as tb_rate, Same as tb_rate, Same as tb_rate, Same as
> > tb_rate},
> > .tc_period = 10,
> > },
> > };
> >
> > [3]
> > static struct rte_sched_pipe_params pipe_profile[] = {
> > { /* Profile #0 */
> > .tb_rate = Link speed (1G/10G..) divided by 8 (bits)/4096 (maximum number
> > of pipes),
> > .tb_size = 1000000,
> >
> > .tc_rate = {pipe's tb_rate, pipe's tb_rate, pipe's tb_rate, pipe's
> > tb_rate},
> > .tc_period = 40,
> >
> > .wrr_weights = {16, 4, 2, 1,  16, 4, 2, 1,  16, 4, 2, 1,  16, 4, 2, 1},
> > },
> > };
> >
> > Our plan here is to initialize the pipe with default profile and modify
> > each pipe based on user configurations.
> > My questions are
> > [a] Can I modify pipe profile during run time using rte_sched_pipe_config
> > ()? (question repeated)
> > If I can modify at pipe level,
> > [b] Can we have different profiles for pipes, With one default profile at
> > initialization?
> > [c] Can we modify port level params without deleting the port using
> > rte_sched_port_config ()?

[a] no
[b] no
[c] no

Read the code.

  reply	other threads:[~2014-10-18  9:09 UTC|newest]

Thread overview: 5+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-10-13 22:54 satish
2014-10-17 21:59 ` satish
2014-10-18  9:16   ` Stephen Hemminger [this message]
     [not found]   ` <251C79C4-74C5-4EF9-A987-3B4DB94A31D5@intel.com>
2014-10-20 19:21     ` Dumitrescu, Cristian
2014-10-20 21:05       ` satish

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=20141018111643.44a9d59f@uryu.home.lan \
    --to=stephen@networkplumber.org \
    --cc=dev@dpdk.org \
    --cc=nsatishbabu@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).