[Spread-users] is spread the right choice ?

John Lane Schultz jschultz at spreadconcepts.com
Tue Apr 3 11:53:12 EDT 2007


Sami M wrote:
>  
> I am wondering if implementing send-side-filtering is a hard problem to 
> solve. Sounds like you'd need to track 'group membership to host 
> mapping' in some sort of local datastructure and filter messages based 
> on that. 
>  

The idea itself is simple, but on closer inspection is more difficult than you 
might think.

For example, Spread currently updates group membership using regular AGREED 
messages sent between the daemons (low overhead + fast).  However, this means 
that if a sending daemon has any holes, then those missing messages might be 
changing the membership to host mapping, which makes it more difficult to do 
correct send-side filtering.

There are plenty of other complications as well, such as what is the optimal 
break point between sending several unicasts and sending one multicast?  The 
daemons do need the meta-state attached with every message, so the meta-state 
will still need to be sent the way whole messages are today.  And plenty more ...

Cheers!

-- 
John Schultz
Spread Concepts LLC
Phn: 443 838 2200
Fax: 301 560 8875




More information about the Spread-users mailing list