*"Is it something you can show?"*

Yes, but not on OCP white box Wi-Fi APs , we are in the process of getting
a NOS (linux) up and running for Edge-Core submitted Wi-Fi AP designs.

These are Broadcom designs and we need a modern kernel 3.x or beyond to
really support network names paces and some advanced networking topologies
using Linux netlink.

These APs don't are storage (Flash) constrained so a stock linux distro
would not work. Probably not even ONL .
We are currently in the process of working towards building a openWRT
distro for these Wi-Fi APs.

Once openWRT is ported (3.18.23 kernel version we are shoot for, as of now)
, user space network automation framework would work without much mods.

We are shooting for a demo in the fall OCP networking workshop, location
TBD.

BR
Rajat Ghai

On Mon, Jun 13, 2016 at 6:52 PM, <jeremy.huylebro...@orange.com> wrote:

>
>
>>    - How large can be the deployments of a CBW?
>>
>> >>>> Large as in per site OR # of sites (multi site L2 / L3 VPNs etc) ?
> Regardless, the scale of the deployments is really based on traffic
> dimensioning and other factors , hard to provide a generic answer ; as such
> there should not be any hard / artificial  limits   .
>
>
> The question was also broad to understand the thinking that was taken into
> account for not only performance and services but also manageability or
> cost.
>
>
>
>>    - What operating system you choose for the switches?
>>
>> >>>>>> The Switches have choice of any NOS ; AFAIK, currently there is
> support for ONL, Cumulus and maybe Ubuntu Snappy (Ubuntu folks can confirm
> availability) ; Going forward We (OCP+Benu) are focusing on tooling and
> instrumentation as well as dynamic network topology management of 10s of
> thousands of such campus/branches.
>
>
> So the idea is to open source entirely the hardware and software
> foundation via OCP? That would be great in this case for our own usage.
>
>
>>    - Is the network team developer enough in its mindset to program some
>>    advanced rules/automation or handle the Linux CLI?
>>
>> >>> Yea.. centrally managed IFTTT style automation for multiple branches
> from a CBW manager in the Data Center.
>
>
> Is it something you can show?
>
>
>>    - Was the architecture and hardware sort of over-provisioned for
>>    future use cases (ex: x86 CPU, more memory, more ports, bigger ports,
>>    fiber)?
>>
>> >>>>> CBW H/W is always price sensitive and hence compute / storage
> constraint. In the current state, you can assume a dual core ARM SoC as a
> compute engine in the White Box APs and Switches. Flash any where from 256
> MB to a few GBs,  DRAM from 128 MBs to 1 GB ; From compute perspective, the
> CBW switches prolly would can run containerized workloads, but it is tough
> to do it on the APs, though we are using LXC on the openWRT NOS on APs. The
> benefit of running containerized workloads on the CBW gear is to facilitate
> automated dark launches w/o destabilizing the networking topology.
>
>
> Definitely interested as well in containerized apps/functions on the
> switches.
>
> _________________________________________________________________________________________________________________________
>
> Ce message et ses pieces jointes peuvent contenir des informations 
> confidentielles ou privilegiees et ne doivent donc
> pas etre diffuses, exploites ou copies sans autorisation. Si vous avez recu 
> ce message par erreur, veuillez le signaler
> a l'expediteur et le detruire ainsi que les pieces jointes. Les messages 
> electroniques etant susceptibles d'alteration,
> Orange decline toute responsabilite si ce message a ete altere, deforme ou 
> falsifie. Merci.
>
> This message and its attachments may contain confidential or privileged 
> information that may be protected by law;
> they should not be distributed, used or copied without authorisation.
> If you have received this email in error, please notify the sender and delete 
> this message and its attachments.
> As emails may be altered, Orange is not liable for messages that have been 
> modified, changed or falsified.
> Thank you.
>
>
_______________________________________________
opencompute-networking mailing list
Unsubscribe: http://lists.opencompute.org/mailman/options/opencompute-networking

opencompute-networking@lists.opencompute.org
http://lists.opencompute.org/mailman/listinfo/opencompute-networking

Reply via email to