[jadmin] s2s only vs scalabe and clustered network

Paolo Perazzo paolo_perazzo at hotmail.com
Thu Oct 27 12:55:18 CDT 2005

I agree with your comment: in fact I mentioned scalable (i.e. s2s) AND 
clustered solution vs s2s only. I wasn't implying on not relying on s2s: the 
mentioned Pampero (Jive) and ejabberd both already supports s2s...
Probably a better way to put this question would have been "what are the 
negative implications of relying on s2s only?"

>From: maqi at jabberstudio.org
>Reply-To: Jabber server administration list <jadmin at jabber.org>
>To: Jabber server administration list <jadmin at jabber.org>
>Subject: Re: [jadmin] s2s only vs scalabe and clustered network
>Date: Thu, 27 Oct 2005 17:48:25 +0200 (CEST)
>On Wed, 19 Oct 2005, Paolo Perazzo wrote:
> > i was wondering what are the difference of implementing a network of
> > multiple servers only talking to each other w/ s2s vs a truly scalable 
> > custered solution that for example project Pampero for Jive or ejabberd 
> > trying to achieve.
>Talking of "truly scalable" for a cluster system is a bit misleading I
>- if there were only clustering solutions and no s2s (say, there's only
>   jabber.org running ejabberd or whatever, but without s2s), the system
>   was not scalable at all in a sense: Nobody apart from the jabber.org
>   admins is able to add new components to the system as all parts of the
>   cluster need to be trusted.
>- this isn't the case with the s2s architecture since servers don't have
>   to trust each other concerning handling of internal data structures
>   (user credentials...) then.
>So s2s and clustering are simply solutions for different problems.
>jadmin mailing list
>jadmin at jabber.org
>FAQ: http://www.jabber.org/about/jadminfaq.shtml

More information about the JAdmin mailing list