axis-java-dev mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Rajith Attapattu" <rajith.attapa...@redknee.com>
Subject RE: Clustering Axis2
Date Fri, 30 Dec 2005 01:14:46 GMT

Sorry for the late reply and I am still feeling drowsy with the pail
killers after my surgery so pardon me if there are any mistakes :)

-----Original Message-----
From: Steve Loughran [mailto:stevel@apache.org] 
Sent: Tuesday, December 20, 2005 5:42 AM
To: axis-dev@ws.apache.org
Subject: Re: Clustering Axis2

>>What do you mean by RSS-based?
Sanjiva mentioned on the list that a RSS based approach could be used to
cluster config data.

What I assume is that each config impl will produce a RSS feed out of
the config info on change and other nodes will consume the RSS feed to
update the local copy.

Please correct me if I am wrong.

>>1. Anubis (under http://smartfrog.org/ ) is what we use for high
>>availability clustering; its an LGPL fault tolerant tuple space. 

Isn't LGPL license a problem when used within apache. I remember for
Geronimo we had to write our own JavaMail impl and couldn't use some
very good open source mail clients due to the fact that they were LGPL
or other type of licensing.

I am no expert on licensing, so can somebody clarify this.

>>2. What does JBoss use for clustering?
They seem to have there own HA-JNDI. There is an effort currently to
merge WADI and Active Space to achieve the same thing within Geronimo.

As far as using it within Axis2 depends on what should be clustered and
at what level.

Sanjiwa has indicated that the deployment interface should be clustered
at the minimum.

Axis2 is more XML centric than Java Object centric, therefore the other
clustering mechanisms that are more object focused may not be suitable
for axis2. (I maybe wrong here please correct me !!!)

>>1. think of what your requirements really are. Separate the minimum
and >>then the 'would like' things.
+1
Steve I totally agree with you point and think there should be more
discussion on this topic as to what should be clustered and how it
should be done.

Regards,

Rajith Attapattu.


Mime
View raw message