[Spread-users] Minimum size message

Gustavo Castro comadreja_98 at hotmail.com
Sat Apr 9 11:33:02 EDT 2005

i should have explained myself better. I´ve done a Primary database backup 
system, which is a distributed database system where the updates can be done 
only in the primary database, and after those changes are done the rest of 
the databases get a message with the instructions to update their data so 
all of them have the same information.
In order to send the queries from secundaries databases to the primary one, 
and to send back the results and updates, i use the Spread multicast 
function. By the way, the database systems works under SQL Server.
While running the tests i got an average of 300 msg per query when working 
against the primary database, and in this case the Spread is not used. 
However, when i did the same tests connecting to the secondary database 
which will use the Spread to send the query and get the answers, time rose 
up to 4000 msg (average). All the computers are the same, so i´ve been 
wondering if this delay is because of Spread waiting to reach a minimum 
message size before sending it (maybe using a timeout), or if the way i send 
the messages doesn´t send them instantly. This is how i do it

SpreadMessage msgres = new SpreadMessage();

Any help would be greatly appreciated, so i can fix any problem or at least 
start looking for the solution in other direction. Either way, thanks!!

>From: Jonathan Stanton <jonathan at cnds.jhu.edu>
>To: Gustavo Castro <comadreja_98 at hotmail.com>
>CC: spread-users at lists.spread.org
>Subject: Re: [Spread-users] Minimum size message
>Date: Sat, 9 Apr 2005 11:13:38 -0400
>The data you send in a SP_multicast call can be as small as 0 bytes (if
>all you want to send is a signal message with a particular mess_type
>The one special case for small messages actually improves performance
>with them because it packs small messages into full-sized packets if you
>send a lot of small messages at once. If you only send one small message
>it is not delayed any more then it would be if it was full-sized (it is
>not like NAGLE in TCP) and it is sent as soon as the flow-control allows
>Can you send more details about the performance problem you notice? For
>example, a description of the scenario or out put from spflooder or
>spmonitor showing the problem.
>On Fri, Apr 08, 2005 at 12:39:57PM +0000, Gustavo Castro wrote:
> > Hello,
> > i´ve been running some tests and some of the times i´ve got are a bit
> > strange. Is there any minimum size for a spread message? I mean, if it´s
> > not big enough the data i´m sending, will spread wait for a timer to 
> > zero, or will it send the message regardless the small size of the data?
> > Thank you!
> >
> > _________________________________________________________________
> > Un amor, una aventura, compañía para un viaje. Regístrate gratis en MSN
> > Amor & Amistad. http://match.msn.es/match/mt.cfm?pg=channel&tcid=162349
> >
> >
> > _______________________________________________
> > Spread-users mailing list
> > Spread-users at lists.spread.org
> > http://lists.spread.org/mailman/listinfo/spread-users
>Jonathan R. Stanton         jonathan at cs.jhu.edu
>Dept. of Computer Science
>Johns Hopkins University

Descarga gratis la Barra de Herramientas de MSN 

More information about the Spread-users mailing list