Re: Standby abuse

  • From: "Niall Litchfield" <niall.litchfield@xxxxxxxxx>
  • To: ddelmoli@xxxxxxx
  • Date: Fri, 7 Nov 2008 21:15:58 +0000

How well does Logical DG deal with tables that don't have primary keys or
unique indexes? How many 3rd party apps are certified to run on a database
that used to be a Logical Stdby? Is even Oracle Apps so certified?  HA and
DR is about more than just a technical data replication solution.

In addition I don't see why the 'advantages' listed below under "efficient
use of resources" are actually advantages of logical stdby over physical
either.

On Fri, Nov 7, 2008 at 8:00 PM, Dominic Delmolino <ddelmoli@xxxxxxx> wrote:

> In an effort to chum the water:
> Why wouldn't DG Logical Standby always be preferable to Physical Standby?
>
> According to the 11g DataGuard Concepts guide:
>
> Benefits of a Physical Standby Database
>
> A physical standby database provides the following benefits:
>
>    -
>
>    Disaster recovery and high availability
>
>    A physical standby database is a robust and efficient disaster recovery
>    and high availability solution. Easy-to-manage switchover and failover
>    capabilities allow easy role reversals between primary and physical standby
>    databases, minimizing the downtime of the primary database for planned and
>    unplanned outages.
>    -
>
>    Data protection
>
>    A physical standby database can prevent data loss, even in the face of
>    unforeseen disasters. A physical standby database supports all datatypes,
>    and all DDL and DML operations that the primary database can support. It
>    also provides a safeguard against data corruptions and user errors. Storage
>    level physical corruptions on the primary database will not be propagated 
> to
>    a standby database. Similarly, logical corruptions or user errors that 
> would
>    otherwise cause data loss can be easily resolved.
>    -
>
>    Reduction in primary database workload
>
>    Oracle Recovery Manager (RMAN) can use a physical standby database to
>    off-load backups from a primary database, saving valuable CPU and I/O
>    cycles.
>
>    A physical standby database can also be queried while Redo Apply is
>    active, which allows queries to be offloaded from the primary to a physical
>    standby, further reducing the primary workload.
>    -
>
>    Performance
>
>    The Redo Apply technology used by a physical standby database is the
>    most efficient mechanism for keeping a standby database updated with 
> changes
>    being made at a primary database because it applies changes using low-level
>    recovery mechanisms which bypass all SQL level code layers.
>
>  Benefits of a Logical Standby Database
>
> A logical standby database is ideal for high availability (HA) while still
> offering data recovery (DR) benefits. Compared to a physical standby
> database, a logical standby database provides significant additional HA
> benefits:
>
>    -
>
>    Protection against additional kinds of failure
>
>    Because logical standby analyzes the redo and reconstructs logical
>    changes to the database, it can detect and protect against certain kinds of
>    hardware failure on the primary that could potentially be replicated 
> through
>    block level changes. Oracle supports having both physical and logical
>    standbys for the same primary server.
>    -
>
>    Efficient use of resources
>
>    A logical standby database is open read/write while changes on the
>    primary are being replicated. Consequently, a logical standby database can
>    simultaneously be used to meet many other business requirements, for 
> example
>    it can run reporting workloads that would problematical for the primary's
>    throughput. It can be used to test new software releases and some kinds of
>    applications on a complete and accurate copy of the primary's data. It can
>    host other applications and additional schemas while protecting data
>    replicated from the primary against local changes. It can be used to assess
>    the impact of certain kinds of physical restructuring (for example, changes
>    to partitioning schemes). Because a logical standby identifies user
>    transactions and replicates only those changes while filtering out
>    background system changes, it can efficiently replicate only transactions 
> of
>    interest.
>    -
>
>    Workload distribution
>
>    Logical standby provides a simple turnkey solution for creating
>    up-to-the-minute, consistent replicas of a primary database that can be 
> used
>    for workload distribution. As the reporting workload increases, additional
>    logical standbys can be created with transparent load distribution without
>    affecting the transactional throughput of the primary server.
>    -
>
>    Optimized for reporting and decision support requirements
>
>    A key benefit of logical standby is that significant auxiliary
>    structures can be created to optimize the reporting workload; structures
>    that could have a prohibitive impact on the primary's transactional 
> response
>    time. A logical standby can have its data physically reorganized into a
>    different storage type with different partitioning, have many different
>    indexes, have on-demand refresh materialized views created and maintained,
>    and it can be used to drive the creation of data cubes and other OLAP data
>    views.
>    -
>
>    Minimizing downtime on software upgrades
>
>    Logical standby can be used to greatly reduce downtime associated with
>    applying patchsets and new software releases. A logical standby can be
>    upgraded to the new release and then switched over to become the active
>    primary. This allows full availability while the old primary is converted 
> to
>    a logical standby and the patchset is applied.
>
>
> Based on this, I see that:
>
> 1. Both provide DR and HA, while Standby has the additional benefit of not
> replicating block-level corruption
> 2. Both can offload backup workload
> 3. Only Standby can be continuously used for reporting and aggregation
> constructs
> 4. Only Standby can be used to support the infamous rolling software
> upgrades
> 5. In theory Standby could handle nologging index rebuilds without
> corruption by skipping all index rebuild DDL
>
> How significant is Physical's performance advantage?
>
> --
> Dominic Delmolino
>



-- 
Niall Litchfield
Oracle DBA
http://www.orawin.info

Other related posts: