Question about an Oracle SID

Question

I'm a bit confused about what a SID is. I understand that it represents a Database instance. However, as I create my database for the first time, I'm asking myself if I really understand what a SID is?
My impression of a SID is a database instance that can contain/manage serveral databases. Is that correct?
However, I wonder if I'm being miss lead by my thinking. During the database creation process it appears that there is a one-to-one relationship between a database and a SID. Meaning that a SID and database are a unique pair and that a one-to-one relationship exist only.
That being the case when it comes the environment variable for "Oracle_SID" does it support having multiple SID's. What is the maximum number of SID that it can support? When connecting to a database in Oracle do I have to supply both the SID and global database name to make a connection? I plan to connect to my databases using both Java and C++.
Thanks

Answer 1

Thanks for you help. However, My question is about
the "Oracle_SID" environment variable. I fully
understand that a SID helps identify resources to
manage a database. However, a server/machine can
have any number of database residing on the
server/machine.
The installation instructions talk about the
environment variables for a machine with the
"Oracle_SID being one of those variables. Based on
what you are telling me a server/machine can have
only 1/one database residing on it based on the fact
that the "Oracle_SID" environment variable can only
accomodate one/1 SID. NO. That is NOT what I said. I said - One SID will equate to One database instance.
You can have any number of SIDs.
BUT you set the SID as an environment variable to tell the software which instance to you wish to connect. ORACLE_SID can only be set to one value. That represents the database instance to which you are about to connect. A tool such as SQLPlus supports only one connection to one instance at a time.
>
Is the "Oracle_SID" environment variable important
when it comes to connecting to a database remotely?Not really. The key is the SERVICE_NAME which is used to tell the LISTENER which SERVICE you want to use.
By default, there is one service that has the same name as the instance's SID.
There is a connection description file, called TNSNAMES.ORA. That file is used the same as /etc/hosts in *nix - pass in an alias and get a connection description - IP in the case of HOSTS; machine, listener port and 'instance information' in the case of TNSNAMES.ORA
The typical connection to an instance is something like
sqlplus hans/test@ORCL
The ORCL is the alias to look up in the TNSNAMES.ORA (or other lookup technique - equivalent to DNS). The client looks up the 3 pieces of information, sends a request to the listener which is (hopefully) listening on the specified port on the indicated machine. It then uses the 'instance information' to ask the database instance for a thread or process.
Oracle8 and older used ORACLE_SID as the 'instance information'
Oracle8i and newer uses SERVICE_NAME as the 'instance information'.
Some Newbie DBAs equate SERVICE_NAME to ORACLE_SID, because one of the services has the same name as the SID. However, one can have up to 64 SERVICES managed by a single instance.
Some Experienced DBAs equate SERVICE_NAME to ORACLE_SID, because they have not spent time in the documentation to understand the difference between SERVICE_NAME and SID. As a result, they generally do not 'get it' when it comes to resource management or instance consolidation.
>
If I'm managing a database at the server and the
server has multiple database residing on it, how do I
define a SID for each database using the "Oracle_SID"
environment variable?If on LINUX, using bash, you simply
export ORACLE_SID=abc
or use the oraenv script to set the SID and all the other environment variables based on the SID.I give a description and example below.
>
I'm just not clear about the relationship between the
"Oracle_SID" environment variable and access a
database when logged onto the database
server/machine.Say I have 2 databases, called 'ORCL' and 'TEST', each with a user (and schema) called 'hans'. Both are down. I want them both up. Therefore I want to start 2 instances.The pattern, on Linux (it's subtly different on Windows) is
# set the environment to start ORCL
. oraenv
ORCL
# or
export ORACLE_SID=ORCL
export ORACLE_HOME=some-directory-path-to-the-ORCL-database's-software
export PATH=$PATH:$ORACLE_HOME/bin
# use the tools to start the database
sqlplus / as sysdba   # note that I did not specify SID ... that was given above
startup
exit
# at this point ORCL database, and all the schemas in the ORCL database, are available
# from anywhere in the network, or on the local machine, I can:
sqlplus hans/password@ORCL
select * from dual;
exit
# but TEST is not up and
sqlplus hans/password@TEST
#  will fail with a 'database not available' type of message
# now set up to start test.  Use oraenv OR do it manually
export ORACLE_SID=TEST
export ORACLE_HOME=the-path-to-the-TEST-database's-software-which-could-be-the-same
export PATH=$PATH:$ORACLE_HOME/bin
# and use the tools to start the database
sqlplus / as sysdba 
startup
exit
# note that I did not specify SID ... that was given above and tells SQLPLUS to establish a connect to the
# instance called TEST.  SQLPlus has special powers in the "as sysdba" mode - it can execute the
# startup process.
# NOW the TEST and the ORCL databases are available through the TEST and ORCL instances
sqlplus hans/password@TEST
select * from dual;
connect hans/password@ORCL
select * from dual;
exit>
I understand that when using the Enterprise Manager I
have to specify the SID and database name but, does
the environment variable "Oracle_SID" have to reflect
a SID for each database that is present on the
server/machine.
The SID tells the system
1) which shared memory segment to use - if we are on the server
2) which instance to connect to over the network, because the SID is the same as the default SERVICE
In Enterprise Manager Database Control, the entry point for the configuration files is a special directory under the ORACLE_HOME that consists of the machine name and the SID. The purpose of setting the ORACLE_SID in this case is to tell the database control which directory to use to find the EM configuration files for that database and instance.
No matter which way you slice it:
1) an Oracle SID is used to point to the configuration information for one database and it's related instance;
2) an Oracle SID is really only meaningful on the machine where the instance is to run;
3) an Oracle SID environment variable can only contain the value for one SID if it is to be intepreted by Oracle programs;
4) an Oracle SID plays no significant part in establishing the connection for an appication;
5) an Oracle SID MAY be used as input to environment-setting programs that make it easier to get the right environment for an application;
6) an Oracle SID is NO LONGER the correct way to set up the TNSNAMES.ORA file to establish a connection.