Accessing business tier cluster using DNS and failvoer1

Question

Hi Guys,
          I think these are jolly interesting questions :-)
          Say I have a web tier and business tier, with business tier and web tier
          each in their own clusters.
          Say I deploy my web applications to web tier, and EJB's to business tier.
          Now if I want to access an EJB from the web tier, I can't just say new
          InitialContext() etc. cos
          I guess the EJB's aren't bound in the web tier, only in the business tier.
          Q1: So do I have to access my business tier by creating a context giving the
          URL of the business tier?
          I am assuming answer to Q1 is yes. If so, the best way is to probably
          somehow map one name to
          each machine. I can do this with DNS easily. Hoever, it is not clear what
          the consequences are
          regarding failover.
          Q2: What if one machine in the business tier goes down? My DNS server will
          still map to it, and hence as DNS round robins if my web tier tries to
          contact my business tier, chances are it may get IP of broken server. What
          will happen? Will it transparently "fail over"?
          Q3: How do I add new machines? Do I have to go and modify the DNS?
          Alternatively, there may be some loadbalancer/router thingie out there that
          I can map by business tier DNS name to, and it
          will transparently route to one of the business servers. I guess they can
          do pings to check health of servers, and possibly see
          new servers coming up?
          Q4: Am I on right track with loadbalancer/router?
          And for the final course :-D, I dont' quite understand something like the
          DataSource which is supposedly clusterable.
          Q5: If I want my business tier to have exclusive access to the database,
          but I want to access a datasource from the web tier, am
          I asking too much? Cos I can't target DataSource to web tier as then I'd
          have to target connection pool, and then I'd have to have
          access to DB from web tier which is somehtin gI want to avoid. ie. is
          there a way to stream data through DS from web tier to business tier.
          If you got this far, thank you!
          Regards,
          Q
          

Answer 1

          Since the EJBs are deployed to a cluster, you can access them with cluster address
          (URL to retrieve context). See cluster examples in your Weblogic installation.
          Your question 5 is quite wierd for me. If you don't want your web tier to touch
          database at all, use EJBs or RMI on business tier to execute your database related
          task for web tier.
          "Quantos Quattro" <[email protected]> wrote:
          >Hi Guys,
          >
          >I think these are jolly interesting questions :-)
          >
          >Say I have a web tier and business tier, with business tier and web tier
          >each in their own clusters.
          >Say I deploy my web applications to web tier, and EJB's to business tier.
          >
          >Now if I want to access an EJB from the web tier, I can't just say new
          >InitialContext() etc. cos
          >I guess the EJB's aren't bound in the web tier, only in the business
          >tier.
          >
          >Q1: So do I have to access my business tier by creating a context giving
          >the
          >URL of the business tier?
          >
          >I am assuming answer to Q1 is yes. If so, the best way is to probably
          >somehow map one name to
          >each machine. I can do this with DNS easily. Hoever, it is not clear
          >what
          >the consequences are
          >regarding failover.
          >
          >Q2: What if one machine in the business tier goes down? My DNS server
          >will
          >still map to it, and hence as DNS round robins if my web tier tries to
          >contact my business tier, chances are it may get IP of broken server.
          > What
          >will happen? Will it transparently "fail over"?
          >
          >Q3: How do I add new machines? Do I have to go and modify the DNS?
          >
          >Alternatively, there may be some loadbalancer/router thingie out there
          >that
          >I can map by business tier DNS name to, and it
          >will transparently route to one of the business servers. I guess they
          >can
          >do pings to check health of servers, and possibly see
          >new servers coming up?
          >
          >Q4: Am I on right track with loadbalancer/router?
          >
          >And for the final course :-D, I dont' quite understand something like
          >the
          >DataSource which is supposedly clusterable.
          >
          >Q5: If I want my business tier to have exclusive access to the database,
          >but I want to access a datasource from the web tier, am
          >I asking too much? Cos I can't target DataSource to web tier as then
          >I'd
          >have to target connection pool, and then I'd have to have
          >access to DB from web tier which is somehtin gI want to avoid. ie.
          >is
          >there a way to stream data through DS from web tier to business tier.
          >
          >If you got this far, thank you!
          >
          >Regards,
          >Q
          >
          >