Test av Freifunk-node i Oslo

[Petter Reinholdtsen]

[Petter Reinholdtsen]
> Jeg legger merke til at Berlin-noden av Freifunk bruker OLSR i stedet
> for batman-adv som rutingsprotokoll.  De nodene som er satt opp i Oslo
> bruker batman-adv.  Kanskje vi burde vurdere å bytte?  Noen som har
> synspunkter?

Jeg spurte på #freifunk (irc.hackint.net) og ble klokere, men ikke vis:


<pere> what is the story behind different communities choice of routing protocol.  berlin seem to use olsr, while others use batman.  why isn't all of freifunk using the same routing protocol?
<pere> I found <URL: http://www.sciencedirect.com/science/article/pii/S089812211100589X > from 2011 claiming olsr perform better than batman.  did this change since then?
<ohrensessel> because there is no "the freifunk network"
<ohrensessel> one of the main ideas of freifunk is given by it beeing decentralized
<ohrensessel> therefore, different local communities --> different choices
<pere> ohrensessel, I guess my question really boil down to this: if one of the protocol is better, why did anyone choose the other?
<ohrensessel> define better ;)
<ohrensessel> batman has the advantage, that it doesn't really need any configuration. e.g. you do not have to assign a unique IP to nodes beforehand as with olsr
<ohrensessel> batman has roaming
<loki-kun> pere: berlin ist using a different software/aproach then most communities in germany
<loki-kun> pere: some use an old protocol like batman-adv ( layer 2) other use olsr/babel (layer3)
<ohrensessel> batman-adv is not old.
<ohrensessel> new versions coming out every few months or so
<loki-kun> ohrensessel: sorry i meant "older" since 2006, but there are newer versions 
<loki-kun> <- still usinh 2013.4 -.-
<ohrensessel> olsr is from 2003 if I'm not mistaken
<loki-kun> pere: you see differnt approaches, different firmwares
<pere> so if we build a new community in norway, what should be prioritize?
<ohrensessel> have a look at the different approaches, especially understand the differences between the olsr and batman-adv (e.g. by having a look at the gluon firmware framework)
<ohrensessel> approach
<pere> no coordinated configuration sound like a good feature.  not quite sure what 'roaming' mean in this regard.
<ohrensessel> roam from one AP to the other
<ohrensessel> in olsr this would mean that the client gets a new IP address
<pere> ah, thought that was the entire point of a mesh network, did not realize olsr did not have it.
<ohrensessel> in berlin they sometimes use batman-adv in local mesh clouds while using olsr to connect to the rest of  the network
<ohrensessel> but batman-adv networks require some central elements: gateways acting as default gateway for the clients and dhcp server
<pere> and olsr do not?
<ohrensessel> as far as I know: no
<ohrensessel> but you need to keep track of the IPs you assigned to nodes to guarantee uniqueness
<loki-kun> pere: olrs is layer 3 so you need a fixed ip afaik if you need english documentation https://wiki.freifunk.net/Freemesh 
<ohrensessel> but this is more an organizational problem, not a technical
<pere> given how few people are involved in oslo, I believe keeping the workload low is a very good idea.
<christiand> but pls don't use olsr, this is old.... u can not use ipv4 and ipv6 at same time... use a nother layer 3 routingprotocol if u would use l3, example: olsrv2, babeld, ospf, bgp...
<ohrensessel> maybe reduce the list to olsrv2 and babel, ospf and bgp are hardly suited for a mesh network I would say
<christiand> i think the idea to use batman for local mesh clouds is a very good idea but layer 3 for long distance radio relay (Richtfunk?) or VPN-tunnel is better
<christiand> we do this at Freifunk in Franken and it is very nice :)
<pere> so why is olsr used in berlin?
<Wonka> historical reasons, I guess.
<christiand> think too
<christiand> and a update is not easy because u must update all router at the same time or need a bridge or so....
<pere> is it not possible to change once local cloud at the time, instead of everything at once?
<pere> of course, it would be possible to deploy new firmware with a timer switch, and switch everyone at the same time using automatic changeover.
<christiand> all clouds are connectet with olsr, if you change one cloud this can not connect to the other clouds... i think this is a problem in berlin but i come not from berlin and don't know exactly
<Wonka> that would need big enough flash in every participating router
<Wonka> with 4MB-routers, that's a problem
<Wonka> <- not from berlin either
<christiand> we have change from olsr to babel... many people must change his gateway, at this time we had a bridge in work to route the old olsr network with the new babel network
<christiand> but now all people have change now und we have babel @work :)
<pere> christiand: why babel and not batman-adv?
<christiand> batman we use too for small "hoods" 
<christiand> but all hoods connectet via Layer 3 and here we use Babel
<christiand> we have about 30 Layer 2 Hoods, in this hoods Batman work
<christiand> all hoods connectet via layer 3 Babel
 cheatha_ chf chris-joki chris1911 christiand christiandr Christian|m 
<christiand> this is a small extract from our network: https://oc.cdresel.de/public.php?service=files&t=27f882c307eee0fedd3f808805ee6906
<pere> christiand: and why isn't batman-adv used across the entire community?  what if two 'hoods' get extended and "meet" each other?
<christiand> Knoten = node ; über = over ; Stadt = city ; Rechenzentrum = data center
<christiand> because freifunk is decentralized, every community do other things
<christiand> if anyone ask me do use our firmware and use our network, we say it is ok, from all over the world (Brazil was the furthest away)
<pere> so just to be different?
<christiand> decentralized ;)

-- 
Vennlig hilsen
Petter Reinholdtsen