© 2009 Oracle Corporation. S311446 : Oracle Database 11g Release 2 First Experiences Plamen Zyumbyulev, Mobiltel Philip Newlan, Oracle RAC Pack.

© 2009 Oracle Corporation

S311446 : Oracle Database 11g Release 2 First Experiences

Plamen Zyumbyulev, MobiltelPhilip Newlan, Oracle RAC Pack

Safe Harbor Statement

The following is intended to outline our general product direction. It is intended for information purposes only, and may not be incorporated into any contract. It is not a commitment to deliver any material, code, or functionality, and should not be relied upon in making purchasing decisions.The development, release, and timing of any features or functionality described for Oracle’s products remains at the sole discretion of Oracle.

Introduction to Oracle Database 11g Release 2 Real Application Clusters

• Features for this session – RAC One Node– Instance Caging– Policy Managed Databases / Server Pools

• New Features not discussed in this session– ASM Improvements and new features• ACFS, AVDM, snapshots, …

– Clusterware Improvements• CTSS, OCR/vote in ASM, …

– Grid Plug and Play (GPnP)– Install Improvements

In 30 mins

RAC One NodeBetter Virtualization for Databases

• A virtualized single instance database • Delivers value of server virtualization

to databases on physical servers– Live migration of instances across

servers– Rolling patches for single instance

databases– Built-in cluster failover for HA– Online upgrade to RAC– Standardized deployment across all

Oracle databasesUse Cases

Load balancing

Database + OS patching and maintenance

Isolation via Instance Caging

• Instance caging enables administrators to limit cpu cores by instance

• Protects service levels by preventing runaway cpu consumption

• Can be adjusted dynamically, while databases are online.– Controlled by cpu_count

parameter– Supports partitioning and

overprovisioning cpu

• Works with Resource Manager

RAC One Node D

cpu_count=1

RAC One Node C

cpu_count=1

RAC One Node Bcpu_count=2

RAC One Node A

cpu_count=4

8 core server

Policy-managed database /Server PoolsNew in Oracle Database 11gR2 • Policy-based management uses server

pools to– Enable dynamic capacity assignment when

needed

– Ensure isolation where necessary (“dedicated servers in a cluster”)

• In order to guarantee:– Applications get the required minimum

resources (whenever possible)

– Applications do not “take” resources from more important applications

Siebel

PSFT

Siebel

PSFT

Ora

cle

Gri

d In

fras

tru

ctu

re

RAC

One

RAC

One

Ora

cle

RA

C D

Bs

FR

EE

FR

EE

RAC

DB2

RAC

DB2

RAC

DB1

RAC

DB1

Resource management with policies

AppApp

Ora

cle

Gri

d In

fras

tru

ctu

re

RAC

One

RAC

One

Ora

cle

RA

C D

Bs

FR

EE

FR

EE

RAC

DB2

RAC

DB2

RAC

DB1

RAC

DB1

Server PoolsNew in Oracle Database 11gR2• logical grouping of servers in the cluster.• Managed by Oracle Clusterware and registered in the

OCR.• One node belongs to one and only one server pool. • Server pools are created and managed using srvctl

commands or through dbcontrol• There are 3 key attributes of a Server Pool– min_size– max_size– Importance

• Default server pool – Free

• Founded in March 1994

• Part of Mobilcom Austria Group

• Telecom leader in Bulgaria

• First production implementation on Oracle 11g GRID

• Best Exadata v1 Performance Results

• First successful Exadata v1 custom installation

• First Customer to run OLTP on Exadata v1

Case Study: M-Tel

M-Tel’s vs. DB Admins’ Challenges

• Provide highest quality of service– Make every application • highly available • scalable • dynamically manage application workload

• Be more flexible– Online maintenance– Service relocation

• Do more with less– Consolidate, simplify, automate

M-Tel’s Grid Architecture 11gR1 2 years ago

Disaster Recovery with real time query

– no idle resources

Business systems

Resource Manager

SOA

Continuous Redo Shipment and Apply

BS 1 BS 2 BS 3 BS 4 BS 7

…

DB1 11g DB2 11gDB3 10g

ASM 11g

CRS 11g

Dedicated and Virtualized EnvironmentsToday

• 40+ Oracle 11g Services consolidated onto 2 Databases and four Linux Servers– 40% Prod Services– 60% Test services

• 25% of all Business Systems are “virtualized” and consolidated– 21% of Prod env– 29% of Test env

DB to Server Consolidation

• Classical approach– Production• 2 DBs per server

– Test• 4+ DBs per server

• 11gR1 GRID Environment– 10+ services per server

M-Tel’s Oracle DB Environment

Resource Manager

SOA

BS 1 BS 2 BS 3 BS 40+…

…

Clusters with more than one DB

Servers running multiple DBs Servers with more

than one DB

Clusters with one DB FO Clusters

Servers with one DB

Problems to further adopt Grid Computing

• Unable to load balance between databases on the same node

• Most applications cannot share the same database (they aren't 100% isolated)

• Some applications aren’t certified with RAC

• No unified approach for single instance database failover. Difficult database movement between servers• No automatic actions in the event of node failure

Oracle Database 11g Release 2RAC ONE – A grid of virtualized single instance databases

DB ADB B

DB CDB D

DB EDB F

• Omotion


DB ADB B

DB CDB D

DB EDB F

• Omotion DB-A from Node 1 to Node 8



DB BDB C

DB DDB E

DB FDB A

DB A

DB B


DB ADB C

DB DDB E

DB F


DB B

• Omotion DB-A from Node 1 to Node 8• Failover DB-B from Node 2 to Node 9

DB ADB C

DB DDB E

DB F


• Omotion DB-A from Node 1 to Node 8• Failover DB-B from Node 2 to Node 9

DB ADB B

DB CDB D

DB EDB F


• Omotion DB-A from Node 1 to Node 8• Failover DB-B from Node 2 to Node 9• Rolling Patches

DB ADB B

DB CDB D

DB EDB F


• Omotion DB-A from Node 1 to Node 8• Failover DB-B from Node 2 to Node 9• Rolling Patches DB-C Node 3 and Node 2

DB ADB B

DB CDB D

DB EDB F


DB C


DB ADB B

DB DDB E

DB F


DB C


DB ADB B

PatchDB D

DB EDB F


DB C


DB ADB B

DB DDB E

DB F


DB C

• Omotion DB-A from Node 1 to Node 8• Failover DB-B from Node 2 to Node 9• Rolling Patches DB-C Node 3 and Node 2• Online Upgrade DB-C to multi-node RAC instances

DB ADB B

DB DDB E

DB F


• Omotion DB-A from Node 1 to Node 8• Failover DB-B from Node 2 to Node 9• Rolling Patches DB-C Node 3 and Node 2• Online Upgrade DB-C to multi-node RAC instances

DB CDB A

DB BDB D

DB EDB F


RAC One NodeContrasting Failover to Omotion – non FAN aware application

• Failover

• Omotion

0

10,000

20,000

30,000

40,000

50,000

60,000

70,000

Failover

60 seconds

0

10,000

20,000

30,000

40,000

50,000

60,000

70,000OMotion

Tra

nsa

ctio

n R

ate

Tra

nsa

ctio

n R

ate

Summary—RAC One Node Benefits:

• Better virtualization for databases– Consolidation, workload management, Better HA– Plus, online maintenance, and DB failover

• Better cluster failover– Additional value (online migration, online maintenance)– Single-vendor solution

• Ready to scale– Online upgrade to multi-node RAC

• Standardized Environment– Common toolset for both single-node and multi-node Oracle

database deployments

Server PoolsWhy ?

• Designed to handle dynamic workloads where you can grow or shrink the pool size as a function of demand, planned or unplanned– Clusters host databases and their respective services in

cluster managed server pools.– Using a cardinality-based approach – nodes become

anonymous.– Server pools are flexible and scalable. You can move servers

between server pools, if needed, based on SLAs– Server pool importance, minimum and maximum number of

servers allowed– Dynamic server pool topology with no down time.

Policy Managed environmentScenario 1.0 - FO

min_size=0

max_size=2

importance=0

min_size=0

max_size=1

importance=0

Nothing happens!!!

Default values!!!

Policy Managed environmentScenario 1.1 - FO

min_size=1

Max_size=2

Importance=0

min_size=1

Max_size=1

Importance=0

min_size=1

Max_size=2

Importance=0

min_size=1

Max_size=1

Importance=0

Policy Managed environmentScenario 2.0 - Reconfigurations

min_size=1

Max_size=3

Importance=0

min_size=0

Max_size=1

Importance=0

Nothing happens!!!


min_size=1

Max_size=3

Importance=1

min_size=0

Max_size=1

Importance=0

Nothing happens!!!


min_size=3

Max_size=3

Importance=1

min_size=1

Max_size=1

Importance=0


min_size=3

Max_size=3

Importance=1

min_size=1

Max_size=1

Importance=1

Nothing happens!!!


min_size=3

Max_size=3

Importance=1

min_size=1

Max_size=1

Importance=2


min_size=2

Max_size=2

Importance=1

min_size=1

Max_size=1

Importance=0

Policy Managed environmentScenario 3.0 - Restart

Sum(min_size)< number of servers

Conclusion

• RAC One Node and policy based managed databases enable us to further improve and extend our consolidation platform

– Better server consolidation

– Enhanced protection from failures

– More flexible and efficient IT environment

– Better Online maintenance

– Better resource isolation

New things in Oracle Database 11gR2 That enable us to move further

• Oracle RAC One Node• Omotion• Failover

• Instance Caging

• Policy Managed Databases/Server Pools

AQ&Q U E S T I O N SQ U E S T I O N S

A N S W E R SA N S W E R S

For More Information

http://search.oracle.com

or

http://www.oracle.com/clusters

Real Application Clusters

© 2009 Oracle Corporation

Oracle RAC One Node – Omotion New in Oracle Database 11gR2

• Omotion moves a Oracle RAC One Node instance from one server to another—without any downtime• Use Cases– Load balancing– Database + OS patching and maintenance

• Oracle supplied tools control migration– Services are not accepting connections on both nodes at the

same time– Migrated instance shutdown transactional once services

moved– A maximum of 30 minutes allowed for connections to migrate

(then shutdown abort)

© 2009 Oracle Corporation. S311446 : Oracle Database 11g Release 2 First Experiences Plamen Zyumbyulev, Mobiltel Philip Newlan, Oracle RAC Pack.

Documents

oracle clusterware

oracle database 11gr2logical

oracle corporations311446

server poolmin

sole discretion of oracle

value of server virtualization

session rac

g release