DPDK patches and discussions
 help / color / mirror / Atom feed
From: "Mcnamara, John" <john.mcnamara@intel.com>
To: "dev@dpdk.org" <dev@dpdk.org>
Subject: [dpdk-dev] FW: OVS with DPDK Meetup notes
Date: Fri, 27 Nov 2015 14:20:26 +0000	[thread overview]
Message-ID: <B27915DBBA3421428155699D51E4CFE2023C747C@IRSMSX103.ger.corp.intel.com> (raw)
In-Reply-To: <BC0FEEC7D7650749874CEC11314A88F74530E8EB@IRSMSX104.ger.corp.intel.com>

Cross posting this from users@dpdk.org to get great visibility.


> -----Original Message-----
> From: dev [mailto:dev-bounces@openvswitch.org] On Behalf Of Traynor, Kevin
> Sent: Thursday, November 26, 2015 5:56 PM
> To: dev@openvswitch.org
> Cc: users@dpdk.org
> Subject: [ovs-dev] OVS with DPDK Meetup notes


 
Hi All,

Just wanted to post some summary notes on the recent OVS with DPDK Meetup we had after the OVS conference. Thanks to everyone for the often lively discussion. 
I've collated and condensed Maryam's notes (Thank you Maryam) with my own. 
Corrections and additions are welcome.

Usability
==========
* Command line options
  - Create defaults which can be overridden on the command line.
  -c 0x1
  -n 4 (this will be in DPDK 2.2 also)
  -socket-mem 1024,0 

* igb / vfio / uio
  - Update documentation to explain these better.

* 1GB v 2MB Hugepages
  - Test to see if significant performance difference between them and if not,
    update documentation to preference 2 MB pages. 

* Assignment of queues to cores/PMD threads
  - Would like to allow for better load balancing of traffic across cores.
  - Add appctl command to get the current queue/core assignments (currently 
    it's only in logs).
  - Discussion about relative merits of trying to do something better than 
    current scheme automatically or allowing a user interface to configure 
    manually. General consensus seemed to be that while it might be difficult 
    to do well automatically, it would be good to investigate and exhaust that 
    possibility before adding a manual user config.
  - Possibility of using DPDK light weight threads discussed.

* Debugging
  - A tcpdump type solution would be nice. Follow DPDK and leverage.

* More flexible port names / auto-discovery of ports
  - Follow DPDK and leverage.

* Single binary for OVS/OVS with DPDK and static vs. dynamic linking
  - Discussion around deployment and what the best model is.
  - Flavio has posted a mail on this
   http://openvswitch.org/pipermail/dev/2015-November/062599.html

* Documentation
  - Make INSTALL.DPDK.md guide cover the most common use cases only, and split 
    the other stuff into a less common/advanced guide. 

Testing
=======
* General agreement in the room that VSPERF pushing a nightly test report for
  OVS/OVS with DPDK to the ovs-build ML would be useful.

* Static Analysis
  - Is there an open source static analysis tool that the community could all use?

Features
========
* Multiqueue vhost-user
  - Looks really promising - will help us scale out performance to the VM. 

* vhost-cuse
  - Still used by some. Give notice to deprecate as there's a maintenance cost. 
    2 releases was mentioned, but I'm not sure that's clear to a user without a 
    regular release cadence. 

* dpdkr/ivshmem
  - Still useful. Check/Update documentation to ensure limitations are clear.


Performance
===========
* Some new optimizations coming in DPDK for vhost performance that we can
  leverage.
* Scalability
  - Performance is scaling well across cores (See Gerald Roger's OVS conf 
    presentation for reference)


At that point everyone was "conferenced out", but tentative plan to arrange another meetup in ~6 months' time if there's enough discussion items to warrant it.

Thanks,
Kevin.

       reply	other threads:[~2015-11-27 14:20 UTC|newest]

Thread overview: 4+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
     [not found] <BC0FEEC7D7650749874CEC11314A88F74530E8EB@IRSMSX104.ger.corp.intel.com>
2015-11-27 14:20 ` Mcnamara, John [this message]
     [not found] ` <20151130235111.GC8825@x240.home>
2015-12-01 11:11   ` [dpdk-dev] [ovs-dev] " Traynor, Kevin
2015-12-02  0:54     ` Loftus, Ciara
2015-12-02  9:22       ` Gray, Mark D

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=B27915DBBA3421428155699D51E4CFE2023C747C@IRSMSX103.ger.corp.intel.com \
    --to=john.mcnamara@intel.com \
    --cc=dev@dpdk.org \
    /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).