Why Clustering

Question

Hi,
          We are in the process in making our decision whether we should
          use clustering or not, to achieve our high scalable need.
          what we are planning is, run weblogic in multi-jvm environment
          without having any knowledge about each other. FYI, we are having
          one dedicated WL server to cater data access layer and data
          caching, which will allow us to achieve persistence integrity. we will be
          using Toplink for WL.
          I have few basic question regarding clustering:
          1) Why should I use clustering ?
          2) What WL clustering buy me ?
          3) What I will loose if I will not use weblogic clustering?
          4) Is Scalability is the only factor people use clustering?
          Any help/pointer is highly appreciated.
          Thanks in advance
          Ahimanikya Satapathy
          OrderCare.Com
          

Answer 1

What you mentioned is something like clustering. It seems you are saying
          that you can just run many boxes/web-logic and that is fine. The problem is,
          if each of those servers has its own non-clustered weblogic running, then
          all the sessions stored on that server get lost if the server dies. Even if
          you can load-balance among many individual non-clustered web-logics, you
          wont be able to "share" the state of the session data amongst the other
          running servers. Clustering makes sure (through memory) that if server A has
          sessions, it looks for a "buddy" server to copy that session information to.
          So it finds server b. Now A and B have the identical http session
          information (assuming we are talking front-end side). If Server A dies,
          server B takes over and thus your clients don't lose everything they had
          entered (talking JSP pages, javabeans, session scope stuff, etc). So, if you
          have a cart system that people on your site add to the cart, or a multiple
          page "wizard like" form that uses the same one bean across the pages (or
          not), if that server dies, their info is still in memory on another weblogic
          server. Even better, if one of the two servers dies, the remaining server
          located another server (if you have 3 or more running) and then replicates
          its session data out to that server..so it automatically "fails over" for
          you incase one dies.
          Hope that helps a little bit.
          "Ahimanikya Satapathy" <[email protected]> wrote in message
          news:[email protected]...
          > Thanks Wei,
          >
          > I think the answer is not that simple and it does not convince me, I can
          > achieve load-balancing,failover, performance, availability,scalability in
          > other means say for example, I can have multiple weblogic running on
          > multiple powerful boxes to achieve all the factors. so how weblogic
          > clustering helps me in this regard, where it stands ?? I would appreciate
          if
          > I would get in depth answer. Also if any body know some good documentation
          > where I can find a good argument for why clustering , will be a great
          help.
          >
          > -- Ahimanikya
          >
          > Wei Guan wrote in message <[email protected]>...
          > >For all your questions, load-balancing, failover, performance,
          > availability,
          > >scalability, etc are the answers.
          > >
          > >--
          > >Cheers - Wei
          > >
          > >
          > >
          > >Ahimanikya Satapathy <[email protected]> wrote in message
          > >news:[email protected]...
          > >> Hi,
          > >>
          > >> We are in the process in making our decision whether we should
          > >> use clustering or not, to achieve our high scalable need.
          > >>
          > >> what we are planning is, run weblogic in multi-jvm environment
          > >> without having any knowledge about each other. FYI, we are having
          > >> one dedicated WL server to cater data access layer and data
          > >> caching, which will allow us to achieve persistence integrity. we will
          be
          > >> using Toplink for WL.
          > >>
          > >> I have few basic question regarding clustering:
          > >>
          > >> 1) Why should I use clustering ?
          > >>
          > >> 2) What WL clustering buy me ?
          > >>
          > >> 3) What I will loose if I will not use weblogic clustering?
          > >>
          > >> 4) Is Scalability is the only factor people use clustering?
          > >>
          > >>
          > >> Any help/pointer is highly appreciated.
          > >>
          > >>
          > >> Thanks in advance
          > >>
          > >> Ahimanikya Satapathy
          > >> OrderCare.Com
          > >>
          > >>
          > >>
          > >
          > >
          >
          >