Re: [Cerowrt-devel] [Babel-users] Lookingfor

Re: [Cerowrt-devel] [Babel-users] Lookingfor

[Dave Taht]

On Wed, Mar 13, 2013 at 3:34 PM,  <jp@nexedi.com> wrote:
> Hi,
>
>> > There is a budget available in a public institution to sponsor a PhD
>> > student who would work on babel.
>>
>> Do you have a topic in mind?  Who would be the supervisor?  What's the
>> deadline?
>
> Everything can be discussed as long as it helps re6st or SlapOS to achieve further resiliency on enterprise applications.
>
> Supervisor would be at ENST.
>
> Current ideas:
> - scalability of babel (1,000,000 nodes network with no backbone)
> - ability for public babel network to survive attacks
> - new metrics
>
> Regards,
>
> JPS.

Scaling dv to a million is something of a challenge.

I would dearly like someone to work with on attempting to implement
new metrics in babel.

Juliusz has described multiple possibilities that seemed promising
that neither he nor I have had time to work on in isolation.

Also fred baker has outlined some interesting ideas for SRC/DST
routing  and address assignment that were just demoed in the ietf
homenet's ospf implementation last week. I think many of the same
ideas and tlvs can be carried over into babel.

Now that I am wrapping up the bufferbloat work

(see cablemodem results at:
http://www.ietf.org/proceedings/86/slides/slides-86-iccrg-3.pdf )

I can also conceive possibilities for wifi congestion and link quality
aware metrics that haven't been tried before...

I hinted at some of them at the MIT talk, "what's wrong with wireless"
http://www.youtube.com/watch?v=Wksh2DPHCDI&feature=youtu.be




>
>
>>
>> -- Juliusz
>>
>>
>> _______________________________________________
>> Babel-users mailing list
>> Babel-users@lists.alioth.debian.org
>> http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/babel-users
>>
>
> _______________________________________________
> Babel-users mailing list
> Babel-users@lists.alioth.debian.org
> http://lists.alioth.debian.org/cgi-bin/mailman/listinfo/babel-users



-- 
Dave Täht

Fixing bufferbloat with cerowrt: http://www.teklibre.com/cerowrt/subscribe.html

Re: [Cerowrt-devel] [Babel-users] Lookingfor

[Benjamin Henrion]

On Sun, Mar 17, 2013 at 5:35 PM, Dave Taht <dave.taht@gmail.com> wrote:
> On Wed, Mar 13, 2013 at 3:34 PM,  <jp@nexedi.com> wrote:
>> Hi,
>>
>>> > There is a budget available in a public institution to sponsor a PhD
>>> > student who would work on babel.
>>>
>>> Do you have a topic in mind?  Who would be the supervisor?  What's the
>>> deadline?
>>
>> Everything can be discussed as long as it helps re6st or SlapOS to achieve further resiliency on enterprise applications.
>>
>> Supervisor would be at ENST.
>>
>> Current ideas:
>> - scalability of babel (1,000,000 nodes network with no backbone)

You will have to change babeld to only consider a slice of the network.

You can run hundreds of babeld inside small openwrt VPSes:

https://github.com/zoobab/openwrt-openvz

You can also use LXC.

>> - ability for public babel network to survive attacks

OLSRd had some encryption key for the packets, don't know for Babel.

About metrics, there was the idea to export infos from the radio layer
(especially from the linux kernel), don't know if this is possible
right now with the latest kernel changes.

--
Benjamin Henrion <bhenrion at ffii.org>
FFII Brussels - +32-484-566109 - +32-2-3500762
"In July 2005, after several failed attempts to legalise software
patents in Europe, the patent establishment changed its strategy.
Instead of explicitly seeking to sanction the patentability of
software, they are now seeking to create a central European patent
court, which would establish and enforce patentability rules in their
favor, without any possibility of correction by competing courts or
democratically elected legislators."

Re: [Cerowrt-devel] [Babel-users] Lookingfor

[Toke Høiland-Jørgensen]

[-- Attachment #1: Type: text/plain, Size: 258 bytes --]

Dave Taht <dave.taht@gmail.com> writes:

>>> > There is a budget available in a public institution to sponsor a PhD
>>> > student who would work on babel.

Sounds like it could be interesting; what's the time frame for getting
the position filled? :)

-Toke

[-- Attachment #2: Type: application/pgp-signature, Size: 489 bytes --]

Re: [Cerowrt-devel] [Babel-users] Lookingfor

[jp]

[-- Attachment #1: Type: text/plain, Size: 612 bytes --]

Hi,

Any time now or later. It can start as an internship then become a PhD. Quite flexible. The problem is the person: should be a good programmer and a fair mathematician (ie. in the sense of being able to least to define an equation to optimize).

Then I can get the position filled and paid.

Regards,

JPS.


> Dave Taht <dave.taht@gmail.com> writes:
> 
> >>> > There is a budget available in a public institution to sponsor a PhD
> >>> > student who would work on babel.
> 
> Sounds like it could be interesting; what's the time frame for getting
> the position filled? :)
> 
> -Toke
> 

Re: [Cerowrt-devel] [Babel-users] Lookingfor

[Juliusz Chroboczek]

> Scaling dv to a million is something of a challenge.

With or without aggregation?

With aggregation, there should be no problem scaling to 500 subnets of
2000 nodes each.

> I would dearly like someone to work with on attempting to implement
> new metrics in babel.

We are working on that.  The smoothed metrics work is a necessary
prerequisite for interesting metrics (which will be intrinsically
unstable), and Matthieu and I have just finished the details of the
delay-based routing.  Unfortunately, we're being delayed by the fact
that our test network is broken right now (our old server has been
decomissioned, and nobody has had time to set up the new one until
last week).

> Also fred baker has outlined some interesting ideas for SRC/DST
> routing  and address assignment that were just demoed in the ietf
> homenet's ospf implementation last week. I think many of the same
> ideas and tlvs can be carried over into babel.

Yeah, I certainly need to look at that at some point.

-- Juliusz

Re: [Cerowrt-devel] [Babel-users] Lookingfor

[Juliusz Chroboczek]

> You will have to change babeld to only consider a slice of the network.

Manual aggregation can be done without changing anything.  On a border
router for the network 192.168.42.0/24, you say

  ip route add unreachable 192.168.42.0/24 proto static

and in babeld.conf,

  out ip 192.168.42.0/24 ge 25 deny

For reliability, you should also add some filtering rules to avoid
leaking local routes into the other subnets.

Automatic aggregation would be possible with just some minor changes
to the implementation (not the protocol), but deaggregation would need
to trigger a (user-configurable) hold time to avoid transient routing
loops.  (That's okay, though, deaggregation should never happen in
a well-designed network -- if you cannot guarantee that your border
routers are stable, you deserve the hold time.)

-- Juliusz