Hi Adriano,
Haven't done the performance benchmarks yet. Ignoring realtime network
latencies, I will conducts tests on 30 peers (as much as my local machine
could handle) and send the reports tomorrow. As you say if the performance
is well under our expectations we need not segment the p2p network.
Thanks,
Narendran.
On May 7, 2012 5:45 PM, "Adriano Marques" <py.adri...@gmail.com> wrote:
> You sure we need to split the network in order to scale?
>
> It is likely that most peers would only connect to a few dozen other
> peers, and won't try to connect to any more than that, because it is enough
> to be part of the network and communicate with all other peers. Did you run
> any tests to benchmark the performance and figure how much it could handle?
>
>
>
> On Sun, May 6, 2012 at 2:00 PM, Narendran Thangarajan <
> narendran.thangara...@gmail.com> wrote:
>
>> Hi all,
>> Currently, the icm-agent (as in "narendran" branch in icm-agent repo)
>> does this. Whenever a new icm-agent node instantiates, the maidsafe part of
>> it searches for potential peers globally. If a peer is already present, it
>> connects to it and becomes a part of the global network. If no peer is
>> present already, then it shuts down itself and starts up as the first
>> bootstrap node so that the future icm-agents could connect to it. Thus
>> currently we have a single P2P network set up and running.
>>
>> But we need to break it up into parts to make it more scalable. There
>> are two ways in which we could do this.
>> 1. Location based : All peers in a single country could be part of a
>> single P2P network. If this is the case, then we need a IP-to-geolocation
>> mapping service in place. Do we have that service already? Another
>> constraint would be to have atleast one super-peer in each country.
>> 2. Super-peers based : Should the P2P network hold constraints on the
>> number of super-peers. Like there should be atleast two super-peers in a
>> single P2P network, so that when one super-peer goes down, all the reports
>> could be forwarded to the secondary super-peer.
>>
>> What could be done for breaking up the P2P network globally?
>>
>> Thanks,
>> Narendran.
>>
>>
>> ------------------------------------------------------------------------------
>> Live Security Virtual Conference
>> Exclusive live event will cover all the ways today's security and
>> threat landscape has changed and how IT managers can respond. Discussions
>> will include endpoint security, mobile security and the latest in malware
>> threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
>> _______________________________________________
>> Umit-devel mailing list
>> Umit-devel@lists.sourceforge.net
>> https://lists.sourceforge.net/lists/listinfo/umit-devel
>>
>>
>
>
> --
> Adriano Monteiro Marques
>
> http://www.openmonitor.org
> http://www.umitproject.org
> http://www.thoughtspad.com
>
> "Don't stay in bed, unless you can make money in bed." - George Burns
>
>
------------------------------------------------------------------------------
Live Security Virtual Conference
Exclusive live event will cover all the ways today's security and
threat landscape has changed and how IT managers can respond. Discussions
will include endpoint security, mobile security and the latest in malware
threats. http://www.accelacomm.com/jaw/sfrnl04242012/114/50122263/
_______________________________________________
Umit-devel mailing list
Umit-devel@lists.sourceforge.net
https://lists.sourceforge.net/lists/listinfo/umit-devel
