Consolidation Method PROS CONS

Virtualization - Complete OS separation. - Overhead associated with

- Allows differing OS versions in running the hypervisor on
each machine. the hardware.
- Allows differing DB versions in - Overhead associated with
each machine. running an entire OS in
- Allows separation of duties if each VM.
different teams need to control - Each OS has to be patched
each VM. and monitored separately.
- The virtual infrastructure can be - Overhead of multiple DBs
used to provide basic high running on a single
availability (HA) functionality. physical server.
- Each DB has to be patched
and monitored separately.
- Someone needs to take on
the responsibility of
learning about and
maintaining the
virtualization
infrastructure.
- Licensing and support of
the virtualization
infrastructure.
Containers - Reduced overhead as only one - Containers do not provide
OS is running on the physical complete OS separation.
hardware. Depending on
- Reduced OS patching as the OS configuration, you may get
is shared between all the bleed-over effects, where
containers. one container can
- Some separation, making each adversely affect the
container "feel" like a separate performance of another.
installation. - The lack of complete OS
- With a separate Oracle separation means there
installation in each container, may be security
each database could run at a implications where
different database version if containers are concerned.
required. - Operating system patches
- The container functionality can affect all containers.
be used to provide basic high - Someone needs to take on
availability (HA) functionality. the responsibility of
learning about and
maintaining the container
infrastructure.
- Overhead of multiple DBs
running on a single
physical server.
- Each DB has to be patched
and monitored separately.
 - Licensing and support of
the container feature. Not
all container solutions are
supported to run Oracle
products.
Multi-Instance - Reduced overhead as only one - No OS separation.
OS is running. - The lack of complete OS
- Reduced OS patching as the OS separation means there
is shared between all the may be security
containers. implications if you make
- Can share a single Oracle silly mistakes.
installation, or have a separate - Operating system patches
Oracle installation per database, affect all instances.
allowing each database to run at - Overhead of multiple DBs
a different database version if running on a single
required. physical server.
- No additional cost or knowledge - If you are using multiple
needed to support a containers Oracle installations, they
or virtualization. will all have to be patched
- Depending on the setup, separately. If you are using
patching and monitoring may be a shared installation, all
simplified. At a minimum, we databases must be kept at
may only need a single Oracle the same version.
installation and a single Cloud - Does not provide any High
Control agent on the server. Availability (HA)
functionality directly, but
Data Guard and Real
Application Clusters (RAC)
can provide this at an extra
cost.
Schema Consolidation - Reduced overhead as only one - No OS separation, making
OS is running. it harder to prioritize
- Reduced overhead as only one performance of specific
DB instance is running. schemas. Resource
- Only a single database Manager cannot control
installation is present. memory usage.
- No additional cost and - The lack of complete OS
knowledge needed to support a separation means there
containers or virtualization. may be security
- Depending on the setup, implications if you make
patching and monitoring may be silly mistakes.
simplified. At a minimum, we - Operating system patches
may only need a single Oracle affect all applications.
installation and a single Cloud - Database patches affect all
Control agent on the server. applications.
- Instance level changes
affect all applications.
- Does not provide any High
 Availability (HA)
functionality directly, but
Data Guard and Real
Application Clusters (RAC)
can provide this at an extra
cost.
- Database recovery and
flashback have to be
planned carefully as all
top-level operations affect
all schemas. This can be
mitigated using tablespace
point in time recovery
(PITR).
Oracle Multitenant - Reduced overhead as only one - The multitenant option is a
OS is running. chargeable Enterprise
- Reduced overhead as only one Edition option.
DB instance is running. - No OS separation, making
- Only a single database it harder to prioritize
installation is present. performance of specific
- No additional cost or knowledge pluggable databases.
needed to support a containers Resource Manager cannot
or virtualization. control memory usage.
- Depending on the setup, - The lack of complete OS
patching and monitoring may be separation means there
simplified. At a minimum, we may be security
may only need a single Oracle implications if you make
installation and a single Cloud silly mistakes.
Control agent on the server. - Operating system patches
- Namespace separation, allowing affect all applications.
you to have public database - Database patches affect all
links, synonyms and even the applications. This can be
same usernames in each PDB. mitigated by using the
unplug/plugin approach to
patches and upgrades, as
described here.
- Instance level changes
affect all applications.
Some initialization
parameters are PDB-
specific, as described here.
- Does not provide any High
Availability (HA)
functionality directly, but
Data Guard and Real
Application Clusters (RAC)
can provide this at an extra
cost.
- Database recovery and
flashback have to be
planned carefully as all
top-level operations affect
all pluggable databases.
This can be mitigated using
PDB point in time recovery
(PITR).