Mailing-List: contact users-help@activemq.apache.org; run by ezmlm
Precedence: bulk
Reply-To: users@activemq.apache.org
Received-SPF: pass (athena.apache.org: domain of lists@nabble.com designates
 216.139.236.158 as permitted sender)
Message-ID: <13848519.post@talk.nabble.com>
Date: Mon, 19 Nov 2007 16:23:49 -0800 (PST)
From: ktecho <ktecho@gmail.com>
To: users@activemq.apache.org
Subject: Re: High Availability and Load Balancing
In-Reply-To: <ec6e67fd0711191243i2262cd1dk84e8e868e1a11e5e@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=us-ascii
Content-Transfer-Encoding: 7bit
References: <13841668.post@talk.nabble.com>
 <ec6e67fd0711191140i598bb805r60352c05ba9ec168@mail.gmail.com>
 <13844122.post@talk.nabble.com>
 <ec6e67fd0711191218g38b96ffdq24cf4d259be0a4c4@mail.gmail.com>
 <13844866.post@talk.nabble.com>
 <ec6e67fd0711191243i2262cd1dk84e8e868e1a11e5e@mail.gmail.com>


I haven't done any performance test by now (besides testing some
sending/receiving stuff in my notebook), but you sure can understand that I
must provide a Scalability Guide to my customer, so even if I can process
"one hundred million messages a second", I need to provide him with a guide
just in case he need "two hundred million messages a second". That's why I
need to fully understand how this works in ActiveMQ, but I found the docs to
be a bit messy. Could be that english is not my native tongue, too.

The machines that runs the brokers should do some more work (some parsing
and inserting into file or BBDD), so it would be nice if we can use all of
them (even if "all" is 2 or 3) as Masters, as using Slaves means something
like calling the solution Active/Pasive, instead of Active/Active.

Could I use two or three (just in case I need them) different Active
(master) brokers using journaled files or database, and telling them that if
one of them fails, the others takes its messages and routes them to
consumers? Or that can only be done with Master/Slave (Active/Passive)
configurations?

Thanks a lot.


James.Strachan wrote:
> 
> On 19/11/2007, ktecho <ktecho@gmail.com> wrote:
>> Something like 860 KB/sec of text encapsulated into JMS messages of about
>> 150
>> bytes each one, coming from 42 machines. The problem is that we need to
>> do
>> some parsing of the messages on the machine, and that the number of
>> clients
>> could potentially grow. That's why we need some brokers to be Master.
> 
> So here's the common misconception. You can have thousands and
> thousands of JMS clients processing messages concurrently on queues.
> However a single message broker (or master/slave cluster) can deal
> with this kind of volume without breaking into a sweat.
> 
> Have you tried doing some performance benchmarking yet?
> http://activemq.apache.org/activemq-performance-module-users-manual.html
> 
> 
>> The only way to provide HA and Load Balancing is to increase the number
>> of
>> server in pairs (master/slave)?
> 
> See this for more background
> http://activemq.apache.org/clustering.html
> 
> -- 
> James
> -------
> http://macstrac.blogspot.com/
> 
> Open Source Integration
> http://open.iona.com
> 
> 

-- 
View this message in context: http://www.nabble.com/High-Availability-and-Load-Balancing-tf4838228s2354.html#a13848519
Sent from the ActiveMQ - User mailing list archive at Nabble.com.