[Spread-users] network configuration

Matthew Gillen mgillen at bbn.com
Mon Jan 29 15:29:27 EST 2007


KRIEGER Nicolas wrote:
> In fact, there are only two machines who have to communicate. If there
> is a failure, the reslut is the same for both configurations. Is there
> any advantage of using one configuration than the other with only two
> machines?

Well, there is some "steady state" traffic that would be avoided by only using
one daemon.  It may have an effect on the predictability of message delivery
times as well (ie the multi-daemon configuration would generally take longer
to deliver messages).

Both of these points are pretty minor though, unless your apps are
time-sensitive to the level of 10's of milliseconds.  There may be some subtle
changes in the protocols that actually get used on the wire, but it shouldn't
be anything you notice (again, unless you're applications are highly sensitive
to time).

It should be pretty easy to test both configurations yourself using wireshark
or tcpdump or some other packet-dump utility.  It shouldn't even require any
application changes to tinker with the daemon config (unless the connect
string is hard-coded in your apps).

Other people may have more detailed knowledge.
HTH,
Matt

> ----- Original Message -----
> From: Matthew Gillen <mgillen at bbn.com>
> Date: Monday, January 29, 2007 5:18 pm
> Subject: Re: [Spread-users] network configuration
> 
>> KRIEGER Nicolas wrote:
>>> Hi,
>>> I would like to know which of the following configurations is the 
>> best.> Is it better to have a spread deamon on each machine where 
>> there must be
>>> a client ? Or only a deamon on one machine and all clients (on 
>> several> machines) access this machine ? All machines are in the 
>> same network.
>>
>> Depends on whether a single point of failure is acceptable to you.
>>
>> Matt
>>
>>
> 
> 
> _______________________________________________
> Spread-users mailing list
> Spread-users at lists.spread.org
> http://lists.spread.org/mailman/listinfo/spread-users





More information about the Spread-users mailing list