Clustering EJB's Across a Wide-Area-Network

Answer 1

Hi Eric,
          If you have a highly reliable database access or replication strategy, I
          feel it is better to rely on that for failover across sites. IMHO.
          Peace,
          Cameron Purdy
          Tangosol, Inc.
          http://www.tangosol.com/coherence.jsp
          Tangosol Coherence: Clustered Replicated Cache for Weblogic
          "Eric Ma" <[email protected]> wrote in message
          news:[email protected]...
          >
          > As we are facing the same decision I am really interested in this thread.
          Are
          > you suggesting that because multicast across a WAN is not as reliable as
          in a
          > LAN environment, so we shouldn't use it? Do you think it is better to
          have multiple
          > LAN-based WebLogic clusters which do not talk to each other, and rely on
          database
          > replication for failover? Thanks for any advice.
          >
          > Eric Ma
          >
          > "Cameron Purdy" <[email protected]> wrote:
          > >>>I am working with a customer that is considering clustering several
          > >>>Stateless EJB's across a Wide-Area-Network and I'd like to find out
          > >if
          > >>>anyone else has successfully implemented such an environment. If so,
          > >what
          > >>>were the implications (performance, reliability, whatever). If not,
          > >perhaps
          > >>>someone can provide an alternative method for accomplishing load
          balancing
          > >>>and failover between the 2 sites.
          > >
          > >> WLS clustering should work fine across WAN., But as you know multicast
          > >> is not a reliable protocol so you should be very careful in setting
          > >up
          > >> the routers which will route the multicast packets across subnets.
          > >As
          > >> long as multicast packets live and sent reliably over the network you
          > >> should be fine.
          > >
          > >WLS can cluster fine across a lightly-loaded relatively-reliable WAN.
          > >You do
          > >need to turn on the ability to multi-cast across the WAN, which is
          usually
          > >a
          > >battle in and of itself. Usually, a separate cluster of WLS is run on
          > >each
          > >end of the WAN, and is not clustered across the WAN. The real question
          > >for
          > >setting up multi-site DR plans with Weblogic is "what are you failing
          > >over
          > >across the sites?"
          > >
          > >For example, do you have replicated databases? Master/slave or both live?
          > >How do you detect site failure? Is the database at a different site with
          > >the
          > >WL servers at their own 2 sites? Usually it's the database that handles
          > >the
          > >multi-site failover, not the WebLogic.
          > >
          > >BTW - if you're using Coherence with WebLogic, in our 1.2.1 release we
          > >have
          > >added a compression network filter that can drastically improve
          throughput
          > >with WAN clustering. It is available starting with 1.2.1 build 86, which
          > >is
          > >planned for availability later today.
          > >
          > >Peace,
          > >
          > >Cameron Purdy
          > >Tangosol, Inc.
          > >http://www.tangosol.com/coherence.jsp
          > >Tangosol Coherence: Clustered Replicated Cache for Weblogic
          > >
          > >
          > >"Kumar Allamraju" <[email protected]> wrote in message
          > >news:[email protected]...
          > >>
          > >
          > >
          > >
          >