You are viewing a plain text version of this content. The canonical link for it is here.
Posted to server-user@james.apache.org by Keith Chew <ke...@jungledrum.co.nz> on 2002/01/31 22:25:41 UTC
avalon, james and failover...
Hi
I have a general question:
How do we implement failover in a java application?
For example, cases like:
- spool manager needs to perform jobs (ie dispatch requests)
- job scheduler needs to performs tasks
- servers need to send messages
When the program shutsdown or the PC goes down, we need another instance in
another PC to take over. I know in Resin they have a "smart" component
(mod_caucho) that knows about the servers. But in application servers like
JBoss, there's no such component, every instance works the same (I think).
Can someone point me in the right direction to look for documentation on:
- failover theory
- different techniques that can be used
or any other thoughts in this area.
I will contribute my findings to the group, so maybe James could incorporate
somekind of failover mechanism.
Keith
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
RE: avalon, james and failover...
Posted by Keith Chew <ke...@jungledrum.co.nz>.
Hi Eric
I am interested in option (2), but would like to know how to implement it.
I know Jboss is using multicast techniques and looking into Jini.
I am looking for a framework on top of these techniques, which I can use for
my application.
Say I am writing a Job Dispatcher server using Avalon and would like to add
failover and load sharing. It would be nice to have it incorporated into
avalon so that I can use it transparently.
So I guess it's really an Avalon area, but since failover involves data
distribution (eg server is suppose to dispatch a group of jobs, the other
instances must know this information), I don't think it's as transparent as
I would like.
I am just looking into this area, in hope to learn more about it and
hopefully contribute to the community.
Keep the thoughts coming in
Keith
-----Original Message-----
From: Eric Weidner [mailto:eric.weidner@ejbsolutions.com]
Sent: Friday, 1 February 2002 11:24 a.m.
To: James Users List
Subject: Re: avalon, james and failover...
Keith,
A couple of typical solutions to this problem are...
1. Network load balancers - Typical hardware solutions that sit in front of
multiple servers and forward traffic on to the servers in different
balancing modes. These will have some form of health check that can be
configured to test if your server is responding. Lookup Alteon Web Switches
by Nortel Networks, BigIP by F5 (www.f5.com), and Arrowpoint (bought by
Cisco I believe). These are industrial strength hardware layer options used
by big hosting companies (i.e. Loudcloud uses Alteons).
2. J2EE applications such as Weblogic will typically implement some form of
clustering technology where several servers share their information back and
forth to achieve scalability and failover. JBoss may include this as well
if it it's development is far enough along.
Hope this helps,
Eric Weidner
Sr. Consultant
EJB Solutions Inc.
----- Original Message -----
From: "Keith Chew" <ke...@jungledrum.co.nz>
To: "James Users List" <ja...@jakarta.apache.org>
Sent: Thursday, January 31, 2002 2:25 PM
Subject: avalon, james and failover...
> Hi
>
> I have a general question:
>
> How do we implement failover in a java application?
>
> For example, cases like:
> - spool manager needs to perform jobs (ie dispatch requests)
> - job scheduler needs to performs tasks
> - servers need to send messages
>
> When the program shutsdown or the PC goes down, we need another instance
in
> another PC to take over. I know in Resin they have a "smart" component
> (mod_caucho) that knows about the servers. But in application servers like
> JBoss, there's no such component, every instance works the same (I think).
>
> Can someone point me in the right direction to look for documentation on:
> - failover theory
> - different techniques that can be used
>
> or any other thoughts in this area.
>
> I will contribute my findings to the group, so maybe James could
incorporate
> somekind of failover mechanism.
>
> Keith
>
>
> --
> To unsubscribe, e-mail:
<ma...@jakarta.apache.org>
> For additional commands, e-mail:
<ma...@jakarta.apache.org>
>
>
>
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>
Re: avalon, james and failover...
Posted by Eric Weidner <er...@ejbsolutions.com>.
Keith,
A couple of typical solutions to this problem are...
1. Network load balancers - Typical hardware solutions that sit in front of
multiple servers and forward traffic on to the servers in different
balancing modes. These will have some form of health check that can be
configured to test if your server is responding. Lookup Alteon Web Switches
by Nortel Networks, BigIP by F5 (www.f5.com), and Arrowpoint (bought by
Cisco I believe). These are industrial strength hardware layer options used
by big hosting companies (i.e. Loudcloud uses Alteons).
2. J2EE applications such as Weblogic will typically implement some form of
clustering technology where several servers share their information back and
forth to achieve scalability and failover. JBoss may include this as well
if it it's development is far enough along.
Hope this helps,
Eric Weidner
Sr. Consultant
EJB Solutions Inc.
----- Original Message -----
From: "Keith Chew" <ke...@jungledrum.co.nz>
To: "James Users List" <ja...@jakarta.apache.org>
Sent: Thursday, January 31, 2002 2:25 PM
Subject: avalon, james and failover...
> Hi
>
> I have a general question:
>
> How do we implement failover in a java application?
>
> For example, cases like:
> - spool manager needs to perform jobs (ie dispatch requests)
> - job scheduler needs to performs tasks
> - servers need to send messages
>
> When the program shutsdown or the PC goes down, we need another instance
in
> another PC to take over. I know in Resin they have a "smart" component
> (mod_caucho) that knows about the servers. But in application servers like
> JBoss, there's no such component, every instance works the same (I think).
>
> Can someone point me in the right direction to look for documentation on:
> - failover theory
> - different techniques that can be used
>
> or any other thoughts in this area.
>
> I will contribute my findings to the group, so maybe James could
incorporate
> somekind of failover mechanism.
>
> Keith
>
>
> --
> To unsubscribe, e-mail:
<ma...@jakarta.apache.org>
> For additional commands, e-mail:
<ma...@jakarta.apache.org>
>
>
>
--
To unsubscribe, e-mail: <ma...@jakarta.apache.org>
For additional commands, e-mail: <ma...@jakarta.apache.org>