incubator-couchdb-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Dave Cottlehuber <d...@muse.net.nz>
Subject Re: Peer-to-Peer Replication
Date Sat, 09 Apr 2011 07:18:54 GMT
On 9 April 2011 19:14, Chris Anderson <jchris@apache.org> wrote:
> I heard from one experienced iOS dev, that the best way to
> interconnect locally is in GameKit.
>
> Chris
>
> On Fri, Apr 8, 2011 at 11:22 AM, Andrew Tunnell-Jones <andrew@tj.id.au> wrote:
>> I've been pondering a subset of this problem as I want to implement
>> opportunistic replication. I'd like to use DNS Service Discovery
>> (often referred to as ZeroConf) and I've already written an Erlang
>> interface to Apple's Bonjour DNSSD API[1] which works on OS X and
>> Linux via Avahi's compatibility layer. The code should be portable to
>> Windows but I've not gone there yet.
>>
>> Slight digression - Apple Bonjour on OS X and Windows can be
>> configured to advertise and browse services over the internet (it'll
>> setup a port forward via NAT-PMP or uPnP if needed/available) via any
>> DNS server which supports DNS Update and has an appropriately
>> configured zone[2], though it works a lot better if the server
>> implements Apple's extensions for real-time updates (DNS-LLQ) and
>> automatic record expiration (DNS-UL). (I've written a CouchDB backed
>> one that does[3].) Avahi's wide-area support is read-only, though
>> their trac install indicates write support is slated for 0.7[4].
>>
>> Back to CouchDB - I haven't worked out the exact mechanics but I'd
>> like to do something along the lines of starting an SSL mochiweb
>> listener and then advertising it using the hash of the SSL cert
>> (self-signed or otherwise) as the service name. As part of the service
>> advertisement, a user-friendly text identifier would be included that
>> can be arbitrarily changed (as the cert hash is the real identifier).
>> The SSL transport would require mutual acceptance of certificates
>> which would be configured by POST'ing a hash to a particular URL. A
>> list of services (hashes) and further information on a given service
>> (including the friendly identifier) would be retrievable via other
>> URLs. Replication would be configured in a similar manner to
>> _replicator with schema being dropped from target/source URLs and a
>> peers hash being used in place of a hostname. Continuous replication
>> would be persisted between restarts and then setup and torn down as
>> services appear and disappear.
>>
>> Keeping in mind this is mostly conjecture at this point, any thoughts?
>>
>> Andrew
>>
>> 1. https://github.com/andrewtj/dnssd_erlang
>> 2. http://dns-sd.org/ServerSetup.html (dated)
>> 3. https://github.com/andrewtj/dnsxd
>> 4. http://avahi.org/milestone/Avahi%200.7
>>
>>> From: Ryan Ramage <ryan.ramage@gmail.com>
>>> To: user@couchdb.apache.org
>>> Date: Thu, 7 Apr 2011 09:40:48 -0600
>>> Subject: Re: Peer-to-Peer Replication
>>> I think we are missing the issue. We do all agree that couch is great
>>> at replicating when it has been wired up with src and dest urls.
>>>
>>> The issue is more around creating a distributed graph management to
>>> handle nodes (couches) in a peer to peer manner. I don't think this
>>> space has been really explored.
>>>
>>> For a local network, there are lots of service discovery protocols
>>> that you could use, something like
>>> http://en.wikipedia.org/wiki/Zero_configuration_networking or
>>> http://en.wikipedia.org/wiki/Universal_Plug_and_Play
>>> but of course this would be outside of what couch does. I think
>>> BigCouch may have something like that built in, but someone from
>>> Cloudant would have to confirm.
>>>
>>> For a more p2p system, this is a much harder problem. For one, you
>>> mentioned the network ports. If you are imagining average people using
>>> this, then you would have to deal with managing port forwarding using
>>> Internet Gateway Device Protocol. At least one couch on the end of a
>>> replication would have to be able to be accessed by http through a
>>> router. You would want a user friendly way to do this.
>>>
>>> Next is managing the graph. This is hard. No help from couch again.
>>> Nodes going up and down, etc.
>>>
>>> It would be fun to see some work done on this. For extra points it
>>> would be cool if it where done in erlang and could be contributed into
>>> the couch core :)
>>>
>>> Ryan
>>>
>>>
>>>
>>>
>>
>
>
>
> --
> Chris Anderson
> http://jchrisa.net
> http://couchbase.com
>

Is there an advantage in using an erlang-based discovery rather than
iOS library - surely linking up with any nearby couch is preferred?

A+
Dave

Mime
View raw message