On 2021-07-21 12:42 p.m., Maneesh Mohanan Pillai wrote:
> Hi Jamal,
>
> The feature flag was enabled for you yesterday when I sent that email. Only
> the people at the lounge at that time needed to refresh their screen to see
> the change, they don't need to keep doing that. Anyone who wasn't in the
> lounge also wouldn't need to refresh their screen.
>
> Can you send through a screenshot of a table edit screen so I can validate
> that the 50 person tables are on for you? It's working on our end.
>
Ok, my misunderstanding then. You meant people at the time you sent
had to refresh their browsers and i thought you meant that is a general
requirement.
thanks for the clarification. We'll let you know if something breaks.
cheers,
jamal
So it turns out we can have 50 people sit on a table at
the lounge.
Unfortunately, once 8 people are seated it seems to add more
if you are rejected because the table is full you need to
refresh your browser. I know that sounds silly - but airmeet
only recently added this extension and is still work in progress.
Maneesh - please correct me if i didnt describe things correctly.
cheers,
jamal
Hi,
Due to unforeseen circumstances, there is a small change of schedule for
tomorrow.
The session "Understanding Linux Network Stack Overheads for
High Speed Networks". See:
https://netdevconf.info/0x15/session.html?Understanding-Linux-Network-Stack…
will now be the third session instead of the first session as was
planned for earlier.
cheers,
jamal
Hi,
Just a reminder of two features on the conference platform
We do have some really nice setup for the lounge: Tables.
Once you login, go to the lounge and you should see several
tables (we have 100 tables setup). You can grab a table
and invite people over.
You can use these tables at any time - although we encourage
people to attend the sessions instead of loitering in the
lounge ;->
We are also open to reserving tables. If you would like to run
a meeting at the break or after or before the conference just
send email to Jeremy and he can create one for you with a title
of your choice so that other people can find it.
For example, there is a table right now called "Happy hour"
The second feature is ability to replay sessions you missed.
Just go into the schedule and you will see a "replay" button
on each session.
cheers,
jamal
Note there is an ongoing poll of whether we should have the 30 minute
break tomorrow after the TC workshop but before switchdev workshop.
We do have a break today - but not one designated for tomorrow.
Please vote - we are still experimenting to figure how best to do
this.
cheers,
jamal
For the Industry Perspectives Panel on the 19th,
we now have the following vendors confirmed to participate:
Xilinx, Broadcom, Pensando, Intel, Netronome/Corigine and
Nvidia.
We will provide more updates in the next while.
cheers,
jamal
Another reminder:
We are experimenting with a feature that will help people
who are busy during the session or are in different time zones
to still be able to catchup.
As long as you are registered you can always login later on and
replay any session you missed.
cheers,
jamal
On 2021-07-12 7:30 a.m., Jamal Hadi Salim via people wrote:
> Just a reminder that the second phase of netdevconf 0x15 starts
> tomorrow, Tuesday to Wednesday.
> This phase constitutes workshops and tutorials.
>
> Workshops on:
> XDP
> IOT Networking
> TC
> Netfilter
> FRR
> and Switchdev Offload
>
> Tutorial on:
> Introduction to time synchronization over Ethernet
>
> See: https://netdevconf.info/0x15/accepted-sessions.html
>
> cheers,
> jamal
> _______________________________________________
> people mailing list
> people(a)netdevconf.org
> https://lists.netdevconf.info/cgi-bin/mailman/listinfo/people
Just a reminder that the second phase of netdevconf 0x15 starts
tomorrow, Tuesday to Wednesday.
This phase constitutes workshops and tutorials.
Workshops on:
XDP
IOT Networking
TC
Netfilter
FRR
and Switchdev Offload
Tutorial on:
Introduction to time synchronization over Ethernet
See: https://netdevconf.info/0x15/accepted-sessions.html
cheers,
jamal
We are happy to announce our first community sponsor:
The IPsec and Network Security Association!
Thank you for your support.
What is the IPsec and Network Security Association?
The aim of the IPsec and Network Security e.V. (see:
https://linux-ipsec.org/) is to build a strong open
source/protocol community to make network security ready
for the next generation. The activities shall extend and
support the work of standardization organizations with
respect to real world implementation and use cases on open
source operating systems.
Network security technologies became increasingly important
over recent years. In the view of professional attacks against
the core network infrastructure happening these days, network
security technologies must be continously developed to make the
network infrastructure futureproof.
IPsec as the major layer 3 network security protocol plays an
important role in securing todays network. However, as networks
continue to become more and more complex, network security
technologies, like IPsec, must keep up with that increasing
complexity.
cheers,
jamal
We are happy to announce that the Industry Perspectives Panel
will be held on July 19th at 5:50 UTC.
For Netdev conf 0x15 we will get the NIC vendor's perspective
on Smart NICs.
Smart NICs are a new emerging type of NICs which have built in
processors alongside rich ASIC-based packet processing pipelines.
The tradition of the Industry Perspectives is to maintain an open
mind so contrarian perspectives are accepted; for example:
A vendor is allowed to say Smart NICs are a bad idea as long
as they back their claim or that they dont support or want
to support Linux within the NIC.
By Linux we mean the Linux kernel network stack of course.
Then the questions is: What are the requirements that Linux needs
to support to turn around any skeptical vendor?
For vendors supporting Linux, what is the format in which the
kernel running within the NIC operates and interacts with the
host? What is offloaded from within the NIC's kernel vs the
host's?
The panel format is a few technical slides from each vendor describing
the features they offer, their differentiation and if they support
Linux the interaction with Linux. This is followed by questions
from the panel and then an open floor for audience Q&A.
We will be providing more information as to which vendors will
be participating in the next few days.
cheers,
jamal
Hello everybody!
I'll have a talk during Netdev 0x15 about how the life with Switchdev
looks like. It'll be at the Switchdev Offload Workshop. And I want to
collect some statistics to share during my talk. If you use switches
with switchdev-based driver, I invite you to fill out a short form. I
will not take much time. There are only 4 simple questions and only
one is required.
https://forms.gle/FZNVFJ5EgkAyzR7DA
PS. If you wonder how people use Switchdev in real life, I welcome you
on my talk! Of course, people already using switchdev are welcome too
- you'll have your bit of sympathy!
We are pleased to announce the tentative schedule
for 0x15. There may be some minor changes
going forward, if any, but the overall theme will remain.
For both registration and schedule for now go to:
https://netdevconf.info/0x15/virtual.html
This time we have good integration with calendar systems
so should be easy to add to google calendar, etc.
Also one nice feature in the Airmeet system this time is
we are trying a setup such that if you miss a session
because you are busy you can login and play the videos.
As usual, once the videos are cleaned up they will be
uploaded to youtube.
cheers,
jamal
We are pleased to welcome NVIDIA (https://www.nvidia.com/) as a bronze
sponsor for Netdev conf 0x15!
NVIDIA is the leader in open networking with end-to-end solutions for
SmartNICs, DPUs, open optics, and accelerated data center switches.
NVIDIA is committed to open-source at all layers of the software and
hardware stack with open networking initiatives like Linux Networking,
SwitchDev, DENT, and the SONiC network operating systems and believes
that open-source has the power to accelerate innovation and benefits
everyone.
cheers,
jamal
Tom Herbert is still on a quest to speed up the network stack[1].
In this talk he introduces the PANDA parser with intention to
replace the venerable Linux kernel Flow Dissector.
Tom says the flow dissector is hard coded, convoluted making
it both hard to extend and hard to maintain.
(yes, he is guilty as well having been one of the originators
of the flow dissector).
PANDA parser is a domain specific parser that lives under
the philosophy of "write once, run anywhere, run well".
Unlike Flow Dissector, a PANDA Parser with metadata extraction
is written in a declarative representation as opposed to imperative
instructions - all in familiar C. It has been shown that while
more flexible, the PANDA parser is more performant than flowdissector.
The PANDA parser may be compiled to different backends, currently
two implemented backends are available: an optimized userspaces C
and an XDP/eBPF one. There is ongoing work on generating a plain
kernel version as well which may be consumed by other part of the
kernel. For any of those 3 backends, the parser definition stays
unchanged.
[1]https://legacy.netdevconf.info/0x14/session.html?talk-BP4-byte-code-for-programmable-protocol-independent-PDU-processing
More info:
https://netdevconf.info/0x15/session.html?Replacing-Flow-Dissector-with-PAN…
cheers,
jamal
PS: Registration is open, see:
https://netdevconf.info/0x15/registration.html
PPS: We are looking to post the schedule today
Marcelo Leitner and Vlad Buslov will discuss turbo-charging the
TC flower control interface and the evolution of the work that
went into it over a period of two years: At least one hundred
kernel patches, many changes to user space and increase in performance
by a magnitude.
To the datapath skeptics who are wondering why optimize the control
path (reminds of the joke from not-to-be-named kernel guy under the
influence "Why do we need userspace at all?"): You do care if you are
constantly churning or accessing the datapath. Example workloads:
- short lived container workloads where you have to add/delete
access control rules,
- telco environments, where efficient rule statistics are critical
billing
- etc.
cheers,
jamal
PS: Registration is open, see:
https://netdevconf.info/0x15/registration.html
PPS: We are looking to post the schedule today