Oracle Database Administration for Microsoft SQL Server DBAs part 33 pps

Deleting here might be useful for backups, but definitely not something
you would want to do with database files. These are the options that are
available with the ASMCMD utility.
Viewing ASM Information
When connected to the ASM instance, some v$ views give information
about the instances connected, disks that might not be part of a disk group,
and files. For example, the v$asm_disk view shows the disks that are
being used by that database instance, and when viewed from the ASM
instance, it will show all of the disks that are discovered. Table 10-2 lists
some of the ASM v$ views.
302
Oracle Database Administration for Microsoft SQL Server DBAs
View Logged in to ASM Instance
Logged in to Database
Instance
v$asm_client
Shows a row for each
database instance using
ASM
Shows a row for the
ASM instance for the
database
v$asm_disk
Shows all of the disks that
are discovered
Shows only the disks in
the disk groups being
used by this instance
v$asm_diskgroup
Shows all of the disk
groups that are discovered

Shows the disk groups
that are available in the
ASM
v$asm_file
Shows the files for each
disk group mounted
Shows the files for the
instance
v$asm_operation
Shows the file for each
long-running operation
executing in the ASM
instance
Shows no rows
TABLE 10-2.
Some ASM
v$
Views
Here are some examples of using SQL*Plus to take a look at the views in
ASM, making sure the environment is set up to log in:
>export ORACLE_SID= +ASM
>export ORACLE_HOME=/u01/oracle/11.2.0/grid
SQLPLUS> select name, state, total_mb from v$asm_disgroup;
NAME STATE TOTAL_MB

DG_DATA01 MOUNTED 349962
DG_DATA02 MOUNTED 349942
SQLPLUS> select name from v$asm_disk;
NAME

ORADATA01
ORADATA02
ORADATA03
ORADATA04
SQLPLUS> select instance_name, db_name, status
from v$ASM_CLIENT;
SQL> select group_number, instance_name, db_name, status
from v$ASM_CLIENT;
GROUP_NUMBER INSTANCE_NAME DB_NAME STATUS

1 DBDEV DBDEV CONNECTED
2 DB01 DB01 CONNECTED
1 +ASM +ASM CONNECTED
2 +ASM +ASM CONNECTED
4 rows selected.
>export ORACLE_SID=DB01
>export ORACLE_HOME=/u01/oracle/11.2.0/database
SQLPLUS> select group_number, instance_name, status
from v$asm_client;
GROUP_NUMBER INSTANCE_NAME DB_NAME STATUS

1 +ASM DB01 CONNECTED
Notice the difference between the view results on the ASM instance and
the database instance. Also, if there are no disks available in v$asm_disk,
this might indicate an issue with the parameter ASM_DISKSTRING or even
permissions on the directories or devices.
Chapter 10: High-Availability Architecture
303
Streams and Advanced Replication
Replication provides copies of data to different servers, and it can be used to

move data. While it isn’t a failover mechanism usually associated with high
availability, it does help ensure that data is available and can provide a way
to selectively pull out the important data.
SQL Server has replication to distribute transactions to a subscriber. You
create the publisher, which can be various tables, and then you can make
subscriptions to the publisher for replicating to another server. The SQL
Server publisher, distributor, and subscriber fill the roles of capture, stage,
and consume or apply. Some setup with a replication administration user
and another database is required.
For replication, Oracle offers Oracle Streams and the Advanced
Replication option. Which one you use depends on your replication
requirements, including what needs to be replicated in your environment.
Oracle Streams
Oracle Streams, included as part of the Oracle database installation,
captures data changes to distribute to another database. The phases of
Streams are similar to the SQL Server publisher, distributor, and subscriber
roles. A user needs to be created to manage the replication, and a
tablespace is also required.
Setting Up Oracle Streams
The Streams administrator user needs the DBA permissions and admin
privilege on the DBMS_STREAMS_AUTH package. Here is an example for
granting the permissions to the user:
SQLPLUS> grant CONNECT, RESOURCE, DBA to streamadmin;
SQLPLUS> begin DBMS_STREAMS_AUTH.GRANT_ADMIN_
PRIVILEGE( grantee => 'streamadmin', grant_privileges => true);
END;
/
SQLPLUS> grant SELECT_CATALOG_ROLE to streamadmin;
SQLPLUS> grant SELECT ANY DICTIONARY to streamadmin;
The parameter configurations for Streams setup are GLOBAL_NAMES=

TRUE, JOB_QUEUE_PROCESS higher than 2, and STREAMS_POOL_SIZE at
least 200MB. A database link is used to connect to the target server, so the
databases do not need to be identical.
304
Oracle Database Administration for Microsoft SQL Server DBAs
Changes for data and objects are captured and replicated. Replication
can be configured for the whole database, schemas, tables, or even tablespaces.
You can set up Streams through OEM, as shown in Figure 10-9, or through
the DBMS packages.
Through OEM, you can also choose to set up downstream capture and
create an advanced queue. Downstream capture collects streams on a
remote database other than the source. The archive logs can be shipped to
where the downstream capture is configured, or the downstream can be a
real-time data capture. The queue is a messaging system that queues up
information to pass it along for other applications or databases to use to
have persistent messages. This is used for distributed applications to
communicate and coordinate processes in an asynchronous manner.
Having the flexibility to implement Streams for just a schema instead of the
whole database allows you to choose which pieces are more highly available.
The replication doesn’t failover the application to a copy of schema, but
provides a place to get the data, at least via a manual connection. The DBMS_
STREAMS_ADM package has procedures for adding schema and table rules,
and setting up the queue table.
Chapter 10: High-Availability Architecture
305
FIGURE 10-9.
Setting up Oracle Streams in OEM
Using Oracle Streams
Oracle Streams uses logical change records (LCRs) for each row of a table
modified. Each LCR has the name of the table changed, old and new values

for any changed columns, and values for the key columns. This information
can be applied to the rows at the destination sites and resolve conflicts if
they arise.
The changes are captured and staged based on the rules of what is to be
replicated. For the capture side, log-based capture pulls the changes out of
the redo logs. Capturing the information from the redo logs minimizes the
overhead on the system and any of the table changes. The tables that are
marked for replication need to log supplemental information in the redo
logs, such as the primary key columns. The log-based capture has a reader
service that reads the redo logs, and then prepares servers to scan the
defined regions from the reader. The filter of the LCRs is based on the rules
and definitions set up to replicate, so only the changes that are needed are
captured. The builder server merges the records from the preparer and then
passes the change to the staging area for processing. Capturing the changes
from the redo logs can come from the log buffer, active redo, and the
archive log files.
Another capture type is synchronous, which captures the changes as
they are happening. This can be used for tables that might be updated often
and are a smaller subset of tables. It captures the DML changes for each row
and converts it to an LCR. The capture of this change is then passed along to
the staging area.
Using rules for publishing and subscribing to the staging area offers
flexibility in the routing of the streams. The staging area with the queues will
even allow the information to be passed to a database that might not have a
network connection to the main database, by passing through another
database that has connections to both.
The consumption of the information is done by the apply engine. The
apply engine detects conflicts and applies automatically captured DML and
DDL changes. Here again, you have the flexibility of using declarative
transformations or user-supplied functions to set up each LCR.

The source database is kept throughout the Oracle Streams processing.
The administrator controls which changes are to be captured. The apply
engine can be customized with PL/SQL procedures and functions, which
306
Oracle Database Administration for Microsoft SQL Server DBAs
can be registered with the Streams administrator. An example for this is to
apply all of the changes except the deletions on a specific table.
Streams has an advisor that will help with performance and monitoring
of the throughput and latency. The advisor looks at each of the areas in the
process: capture, stage, and apply.
Advanced Replication
Along with Oracle Streams replication, Oracle offers an Advanced
Replication option. This handles master replication with a single master
or multiple masters.
Multimaster replication is known as
peer-to-peer
, and any of the servers
can be updated. Advanced Replication processing to multiple masters can
be asynchronous and synchronous.
For this type of replication, you need to set up a replication admin user.
Tables in the databases should have primary keys. The DBMS_REPCAT
package provides routines for administering and updating the replication
catalog.
Advanced Replication offers the option of replicating to non-Oracle
databases. This allows a way to provide data to several different systems.
Also, with this type of replication, the Oracle database version and platform
do not need to be the same for replication. Advanced Replication may be
suitable for distributed or data warehouse databases, to have copies
available for other systems or to maintain the workload on different servers.
Summary

The role of the DBA is to provide a reliable database environment.
Businesses are requiring that systems be up and available 24/7.
The DBA needs to understand the options for the database system to
provide high availability. This includes knowing the resource costs for
each solution and what type of availability it provides.
Making a case for how a high-availability solution improves the
management and reliability of the database, as well as provides for the
needs of the business, is the responsibility of the DBA. Implementing,
deploying, and administering the environment is the fun part for the DBA.
Chapter 10: High-Availability Architecture
307
SQL Server has high-availability solutions that you might have considered
and/or implemented. The options for SQL Server do not behave the same
as those for Oracle. Even though both have clustering, there are significant
differences. Looking at the differences is a good place to start to understand
the Oracle solutions. However, you also should examine the specific features
and possible configurations of Oracle RAC, Data Guard, and other high-
availability options. Look into the Maximum Availability Architecture for
Oracle, which combines solutions to handle different requirements and
reduce the risks for failures.
308
Oracle Database Administration for Microsoft SQL Server DBAs
appendix
Mental Preparedness
for Multiple Platforms
M
y native language is English, but for some crazy reason,
I wanted to learn German in high school. I learned about
the language, discovered new words, and learned how to
speak a couple of phrases. I started to learn about the syntax

and gender of words. I was able to understand a very basic
conversation, and I could read a simple article. My translations were straight
word for word; I did not understand the slang.
At the university, I was exposed to more German. I studied the grammar
and learned more about the culture, which helped me in understanding
different meanings of phrases instead of just translating each word into English.
Then I lived in Germany for several years, immersed in the language. I started
to be able to speak the phrases and dialect. I even began to dream in German,
and I realized I finally had made the German language a part of who I am.
Some of you might have experienced similar situations with learning a new
language. We also experience this with computer languages and database
platforms.
We have spent this book going through typical tasks and syntax of how
to do things in Oracle, even though SQL Server is your native language. We
used this as a baseline to understand the concepts of the database environment
and tasks that need to be done. The translations were done into Oracle—
some more direct than others. For example, the terms
database
and
instance
were matched up with each other. The database/instance comparison was
to illustrate the Oracle environment—its objects and behaviors—and to truly
convey what Oracle DBAs mean when they say “database.”
Mental preparedness for multiple platforms is being able to distinguish
between the different platforms. It is being able to speak the language of one
platform one minute, and switch over to the language of another platform in
the next minute. You start by learning enough of the differences to make the
transition easier, but then begin to learn about the platform in more depth. It
is still possible to translate between the different platforms, but in doing so,
you might miss the real meaning or underlying concept. Take that extra step

to try to “dream in Oracle.” Try to stop comparing it with SQL Server, and
start embracing some of the Oracle features for what they are.
This is definitely not an easy task. It might take a few years for you to be
able to bounce between platforms without thinking about it. A good place
to start is with some checklists for each platform to help smooth the transition.
310
Oracle Database Administration for Microsoft SQL Server DBAs
Each of the platforms has a different starting point for troubleshooting a
problem. Start a checklist for troubleshooting the problems based on each
platform. It will increase the experience you have with the different database
environments when looking at the issues from different perspectives. How do
you go about looking at connection issues? What about performance? How
about permissions and statistics? Consider where all of these types of issues
fall as database administration activities—daily versus weekly, automatic versus
manual, cause of an issue versus fix for an issue, and so on.
Resolving a performance issue in SQL Server might start with looking for
locks and long-running queries, which could lead to needing to rebuild
indexes. In the Oracle environment, the search might start with looking at
session waits and checking if there are table scans because of stale statistics.
These starting points are good to have for each environment to be able to jump
into a different platform quickly. In not thinking about which platform, quite a
few times I have gone down a path in SQL Server to research a performance
issue by looking at statistics and indexes instead of looking at sp_who2 for
blocking issues. In SQL Server, blocking issues might be higher on my list of
things to check, but I might have skipped a quick check because I didn’t think
of the environment and may have even started to work on tuning the queries.
On the other hand, even though that might not have been the solution, tuning
queries is always worth it for better performance no matter what database
environment. At some point, the troubleshooting may look the same or
converge, depending on the issue, but understanding how the system behaves

and what some of the main issues can be is a good place to start.
Of course, some of the DBA tasks in the environments are the same. For
example, backups are important in any database system, but what is backed
up and the options are different. That could even be said for different databases
in the same platform. Not only are you translating to a different database
platform, but you also must use different applications, which, as a DBA,
you should be used to by now.
For maintenance jobs and tasks, be sure to look at the version of the
database, the platform, and any new features that might make the task easier
or even obsolete. Oracle did this with tablespaces. It was very typical to
coalesce and reorganize the tablespaces and tables, but with locally managed
tablespaces, this became less of a worry. SQL Server might have a higher
priority to rebuild cluster indexes, where Oracle might be looking at statistics
on the tables and indexes for the optimizer to have the correct information.
What is new in the database, the tasks, and maintenance should be reviewed
Appendix: Mental Preparedness for Multiple Platforms
311