Oracle FAQ Your Portal to the Oracle Knowledge Grid
HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US
 

Home -> Community -> Usenet -> c.d.o.server -> Deploying 11i and 10g in a grid environment

Deploying 11i and 10g in a grid environment

From: Equis Uno <ir4u4_at_yahoo.com>
Date: 6 Mar 2004 13:41:28 -0800
Message-ID: <692feddd.0403061341.321add77@posting.google.com>


Hi,

Please respond to
comp.databases.oracle.misc

This is cross posted to c.d.o.s since some of the questions relate to the server.

I'm curious about the challenges a DBA faces when deploying 11i in a grid environment.

Once I get a proper mental handle on what the challenges are, I'd obviously like to find some solutions.

So, assume I have 10 nodes running Linux and that I have the latest version
of 11i and 10g.

Lets assume I decide on this physical deployment:

DB Tier:
3 linux nodes running 10g RAC

Apps Tier:
7 linux nodes running 11i

Question 0:
How sensitive is 11i (11.5.9) to the version of the underlying database?
Will it run fine on 8i, 9i, 10g?  

Question 1:
During 11i installation, when I run rapidwiz, is it possible to point rapidwiz at a database service rather than a specific Oracle instance?

q2: (a generalization of q1)
Is rapidwiz RAC aware?

q3:
Ignoring the answers to q1,2, is it possible to configure 11i so that it
is unaffected by a crash of any instance in the 10g RAC?

q4:
What is the best way to install the 11i software on the 7 Linux nodes? Should I run rapidwiz 7 times such that I have 7 copies of the 11i software
on the local disk drives of the 7 Linux nodes? If I do this, will the 11i software on node2 be identical to the 11i software on say node7? If there is a difference, what is it and how complicated is it?

q5: (a variation of q4)
Suppose I don't want to run rapidwiz 7 times; could I run rapidwiz once on a central file server and then serve (using NFS maybe) the 11i software to the 7 Linux nodes? If I do this, do I need to worry that the deployed software will function only on its orginal node? If it does function only on the original node, would it be possible to use soft links, environment variables, and properly edited configuration files to properly serve the 11i software to the 7 Linux nodes so that it functions on the 7 Linux nodes?

q6:
Assume I can figure out how to properly serve 11i software to the 7 nodes using NFS; what is the best way to protect the application from crash of the file server?

q7:
Lets assume that I have figured out an elegant way to deploy the 11i software and I have it properly configured to function with 10g RAC,
what is the best way to deploy services across the 7 nodes? Should I maybe put the Forms server on nodes1,2,3 and the SelfService server
on nodes4,5,6 and the CM on node 7?
Or, should I put all the services on all 7 nodes so that each node is identical?

q8:
Assume I have a good solution to q7, how do I protect the application from
a crash of one of the 7 nodes? For example, if a user is interacting with
some business object (entering an order for example), and this user's session
is connected to node5 and node5 then crashes, is it possible for 11i to
be configured so that this user session would be seamlessly failed over
to another node?

Sorry about posting so many questions; I've not found any obvious answers
in the documentation.

-moi Received on Sat Mar 06 2004 - 15:41:28 CST

Original text of this message

HOME | ASK QUESTION | ADD INFO | SEARCH | E-MAIL US