Re: [jgroups-dev] DAISYCHAIN

Bela Ban <[email protected]>
Newsgroups gmane.comp.java.javagroups.devel
Message-ID <[email protected]>

Mike Jensen wrote:
> What if one of the nodes in the chain is slower than the other nodes?

That's not an aspect that's handled by the transport (and I consider 
DAISYCHAIN as logical part of the transport); it is flow control (FC) 
that ensures a slow member won't get overloaded. Or, if the member is 
constantly too slow, then it will get excluded by failure detection (FD).

In general, JGroups progresses at the pace of the slowest member until 
that member is excluded.

>   Is it possible to use a topology like this and avoid it from having 
> to run at the speed of the slowest node?

Not in my current design, no. The use case for JGroups has always been 
clusters of servers that are located in the same server room, connected 
to the same switch. All the servers run at the same speed. That's where 
DAISYCHAIN comes in.

We're only starting to look into geographic distribution...

> First thought that comes to mind is send messages in both 
> directions...so that you can route around the slowest node.  But that 
> too does not full solve the problem.

That was the original design of the work at EPFL. However, performance 
wasn't as good as with a single ring.

-- 
Bela Ban
Lead JGroups / Clustering Team
JBoss


------------------------------------------------------------------------------
This SF.net email is sponsored by 

Make an app they can't live without
Enter the BlackBerry Developer Challenge
http://p.sf.net/sfu/RIM-dev2dev 
_______________________________________________
Javagroups-development mailing list
lmpx.com only provides a reader for public news (NNTP) servers. It is not affiliated with the servers or forums shown here and is not responsible for the content of articles, which is written by their respective authors.