presentation - oracle exadata architecture overview
DESCRIPTION
Presentation - Oracle Exadata Architecture OverviewTRANSCRIPT
-
Oracle Exadata Storage Server Architecture
Overview
-
Agenda
Exadata Software Architecture
Exadata Storage Layout
Exadata Storage Scale-O
ut Architecture
2008 Oracle Corporation Proprietary and Confidential
2
-
ExadataSoftware Architecture
2009 Oracle Corporation Proprietary and Confidential
3
-
Exadata Architecture
DB Server
DB Instance
DBRM
ASM
Single-Instance
Database
RAC
Database
DB Server
DB Instance
DBRM
ASM
DB Server
DB Instance
DBRM
ASM
iDB Protocol over
Enterprise
Manager
2008 Oracle Corporation Proprietary and Confidential
4
OEL
CELLSRV
MS
RS
IORM
Exadata Cell
InfiniBand with
Path Failover
InfiniBand Switch/Network
Cell
Control
CLI
OEL
CELLSRV
MS
RS
IORM
Exadata Cell
OEL
CELLSRV
MS
RS
IORM
Exadata Cell
-
Exadata Storage Server Overview
Exadata Storage Server
Data
Data
Data
2008 Oracle Corporation Proprietary and Confidential
5
Network Fabric
Evaluate Disk, CPU, and Network Throughput Standalone
-
Exadata Storage Server Overview
Data
Data
Data
Database Server
RDBMS/ASM instance
SGA
IO Client
IO Layer
ASM layer
Exadata Storage Server
2008 Oracle Corporation Proprietary and Confidential
6
Network Fabric
ASM layer
?
How do we connect the IO/ASM layer of ASM/DB to Exadata?
?
-
Exadata Storage Server Overview
Database Server
RDBMS/ASM instance
SGA
cellsrv
Data
Meta data
Data
Meta data
Data
Meta data
IO Client
IO Layer
ASM layer
Exadata Storage Server
2008 Oracle Corporation Proprietary and Confidential
7
Network Fabric
libcell
ASM layer
libcell linked with DB/ASM talks to cellsrv
iDB protocol is born
Multiple threads in cellsrv
Threads perform
asynchronous IO to disks and network
-
Exadata Storage Server Overview
Database Server
RDBMS/ASM instance
SGA
Data
Meta data
Data
Meta data
Data
Meta data
IO Client
IO Layer
ASM layer
Exadata Storage Server
cellsrv
2008 Oracle Corporation Proprietary and Confidential
8
Network Fabric
libcell
ASM layer
Smart Scan enabled in RDBMS and CELLSRV
-
Exadata Storage Server Overview
Database Server
RDBMS/ASM instance
SGA
Data
Meta data
Data
Meta data
Data
Meta data
/etc/oracle/cell/network-config
IO Client
IO Layer
ASM layer
Exadata Storage Server
cellsrv
2008 Oracle Corporation Proprietary and Confidential
9
Network Fabric
cellinit.ora
Local IP
/etc/oracle/cell/network
libcell
ASM layer
cellinit.ora decides which network takes Storage traffic
-
Exadata Storage Server Overview
Database Server
RDBMS/ASM instance
SGA
cellip.ora
Cells
Data
Meta data
Data
Meta data
Data
Meta data
IO Client
IO Layer
ASM layer
Exadata Storage Server
cellsrv
/etc/oracle/cell/network-config
2008 Oracle Corporation Proprietary and Confidential
10
Network Fabric
cellinit.ora
cellip.ora
Local IP
libcell
ASM layer
cellip.ora on database/ASM host maintains list of cells
New cells can be added to cellip.ora dynamically
/etc/oracle/cell/network
-
Exadata Storage Server Overview
Database Server
RDBMS/ASM instance
dskm
diskmon
SGA
cellip.ora
Cells
Data
Meta data
Data
Meta data
Data
Meta data
IO Client
IO Layer
ASM layer
Exadata Storage Server
cellsrv
/etc/oracle/cell/network-config
2008 Oracle Corporation Proprietary and Confidential
11
Network Fabric
cellinit.ora
cellip.ora
Local IP
libcell
ASM layer
Master diskmon (diskmon) starts with CSS and communicates with cellsrv
Slave diskmon (dskm) is part of every instance and communicates with m
aster diskmon
Handles cell failures, IO fencing, IO resource m
anagement plan propagation
/etc/oracle/cell/network
-
Exadata Storage Server Overview
Database Server
RDBMS/ASM instance
dskm
diskmon
SGA
cellip.ora
Cells
ms
cellcli
Data
Meta data
Data
Meta data
Data
Meta data
IO Client
IO Layer
ASM layer
Exadata Storage Server
cellsrv
/etc/oracle/cell/network-config
2008 Oracle Corporation Proprietary and Confidential
12
Network Fabric
cellinit.ora
cellip.ora
Local IP
libcell
ASM layer
Cellcli allows user interaction and configuration
Management Server (M
S) displays and m
anages creation and deletion of
griddisks, changes in hardware, SNMP traps, alerts, email, m
etrics etc
/etc/oracle/cell/network
-
Exadata Storage Server Overview
Database Server
RDBMS/ASM instance
dskm
diskmon
SGA
cellip.ora
Cells
ms
cellcli
Data
Meta data
Data
Meta data
Data
Meta data
IO Client
IO Layer
ASM layer
Exadata Storage Server
cellsrv
/etc/oracle/cell/network-config
2008 Oracle Corporation Proprietary and Confidential
13
Network Fabric
cellinit.ora
cellip.ora
Local IP
rs
libcell
ASM layer
Restart Server (RS) monitors CELLSRV and M
S, Backup RS m
onitors core RS
RS monitors for process aliveness, memory usage etc
/etc/oracle/cell/network
-
Exadata Storage Server Overview
Database Server
RDBMS/ASM instance
dskm
diskmon
SGA
cellip.ora
Cells
ms
cellcli
Data
Meta data
Data
Meta data
Data
Meta data
IO Client
IO Layer
ASM layer
Exadata Storage Server
cellsrv
/etc/oracle/cell/network-config
2008 Oracle Corporation Proprietary and Confidential
14
Network Fabric
cellinit.ora
cellip.ora
Local IP
rsADR
adrci
libcell
ASM layer
Trace files and alert logs in Automatic Diagnostic Repository on cell
alert.log (from RS and CELLSRV), m
s-odl.log, ms-odl.trc, rs*trc, svtrc*.trc
/etc/oracle/cell/network
-
Exadata Storage Server Overview
Database Server
RDBMS/ASM instance
dskm
diskmon
SGA
cellip.ora
Exadata cells
ms
cellcli
Data
Meta data
Data
Meta data
Data
Meta data
IO Client
IO Layer
ASM layer
Exadata Storage Server
cellsrv
/etc/oracle/cell/network-config
2008 Oracle Corporation Proprietary and Confidential
15
Infiniband Fabric
cellinit.ora
cellip.ora
Local IP
rsADR
adrci
libcell
ASM layer
Infiniband fabric requires RPMs from OFED/OEL, RedHat 5.1, OEL 5.1
Exadata Storage Server works only with 11.1.0.7 Database/ASM
/etc/oracle/cell/network
-
Exadata Storage Server Overview
From the Enterprise
ms
cellcli
Data
Meta data
Data
Meta data
Data
Meta data
Enterprise M
anager
OMS
Agent
dcli
Exadata Storage Server
cellsrv
2008 Oracle Corporation Proprietary and Confidential
16
Network Fabric
rsADR
adrci
EM Plugin Provides central location for metrics and alerts across cells
No agent runs on cell
dcli allows user to run commands across cells
ssh/cellcli
-
ExadataStorage Layout
2009 Oracle Corporation Proprietary and Confidential
17
-
Exadata Storage Layout
Physical disks map to a Cell Disks
Cell Disks partitioned into one or multiple Grid Disks
ASM diskgroups created from Grid Disks
Transparent above the ASM layer
2008 Oracle Corporation Proprietary and Confidential
18
Cell
Disk
Sys Area
Sys Area
Grid Disk n
Grid Disk 1
ASM disk
ASM disk
Physical
Disk
-
Exadata Cell
Exadata Cell
Exadata Storage Layout Example
Cell Disks
Cell
Disk
2008 Oracle Corporation Proprietary and Confidential
19
Cell Disk is the entity that represents a physical disk
residing within a Exadata Storage Cell
Automatically discovered and activated
-
Exadata Cell
Exadata Cell
Exadata Storage Layout Example
Grid Disks
Grid
Disk
2008 Oracle Corporation Proprietary and Confidential
20
Cell Disks are logically partitioned into Grid Disks
Grid Disk is the entity allocated to ASM as an ASM disk
Minimum of one Grid Disk per Cell Disk
Can be used to allocate hot, warm
and cold regions of a
Cell Disk or to separate databases sharing Exadata Cells
-
Exadata Storage Layout Example
ASM Disk Groups and Mirroring
Exadata Cell
Exadata Cell
Hot
Hot
Hot
Hot
Hot
Hot
Hot ASM
Disk Group
Cold ASM
Disk Group
2008 Oracle Corporation Proprietary and Confidential
21
Two ASM disk groups defined
One for the active, or hot portion, of the database and a
second for the cold or inactive portion
ASM striping evenly distributes I/O across the disk group
ASM mirroring is used protect against disk failures
Optional for one or both disk groups
Hot
Hot
Hot
Hot
Hot
Hot
Cold
Cold
Cold
Cold
Cold
Cold
-
Exadata Storage Layout Example
ASM Mirroring and Failure Groups
Exadata Cell
Exadata Cell
Hot
Hot
Hot
Hot
Hot
Hot
ASM
Failure Group
ASM
Failure Group
2008 Oracle Corporation Proprietary and Confidential
22
ASM mirroring is used protect against disk failures
ASM failure groups are used to protect against cell failures
Hot
Hot
Hot
Hot
Hot
Hot
Cold
Cold
Cold
Cold
Cold
Cold
ASM
Disk Group
-
ExadataScale-Out Architecture
2009 Oracle Corporation Proprietary and Confidential
23
-
Scale-Out Architecture
The database, ASM, and Exadata
Cells each play a role in Oracles
scale-out storage architecture
Responsibilities are placed in the
optimal location
DB Business Data Protection
ASM
Reliable Storage
Business
Data Protection
2008 Oracle Corporation
24
DB Business Data Protection
ASM Reliable Storage Pool
Exadata Cell Database Intelligent
Storage
Seamless integration simplifies
management
Reliable Storage
Pool
Exadata
Database
Intelligent
Storage
-
Exadata Scale-Out Overview
Cells implement Oracles iDB data protocol
Expose storage as collections of intelligent
network disks called Grid Disks to ASM and the
DB
iDB is conceptually similar to iSCSI but has
extensive DB intelligence
iDB is layered on top of ZDP network protocol
Grid Disks
2008 Oracle Corporation
25
iDB is layered on top of ZDP network protocol
To ensure full scale-out benefits, cells
never communicate with each other
Cross cell operations are implemented in ASM or
DB
Cell independence ensures no perform
ance
bottlenecks, and no cascading failures
Key to scalability architecture
iDB data protocol
ZDP net protocol
-
Exadata Cell Integration for Scale-Out
Cell appliance design elim
inates storage configuration
missteps and administrative overhead
Simple provisioning
Cell grid disks are automatically made visible to ASM
No OS level LUNs or mount points to setup and manage
Cross cell ASM m
irroring is automatically configured
2008 Oracle Corporation
26
Cross cell ASM m
irroring is automatically configured
for grid disks
ASM failure group topology
Multiple Grid disks per physical disk allow m
ultiple
ASMs to use the same cells
-
Automatic Storage Management
Integrated cluster volume
manager
Flexible data distribution
(striping)
Mirroring
ASM Disk
ASM Disk
ASM Disk
ASM Disk
ASM Disk
2008 Oracle Corporation
27
ASM creates a reliable storage pool
from Exadata Grid disks
Mirroring
Automatic data re-balancing
Free
-
ASM Scale-out Data Distribution
ASM manages storage in megabyte
allocation units
Each DB file consists of a set of
allocation units
The location of a files allocation units
are individually tracked by ASM
ASM evenly spreads allocation units
DB
File #1
DB
File #2
Files automatically spread
across disks to balance load
ASM
2008 Oracle Corporation
28
ASM evenly spreads allocation units
across all cells and disks in the grid
All disks evenly utilized
Optimal perform
ance
File #1
File #2
-
ASM Data Re-Distribution
Cells can be easily added or removed
from the storage grid
ASM migrates a fraction of the allocation
units to the new cell
Online and transparent to application
Minimal data m
ovement to new or
removed cells
DB
File #1
DB
File #2
ASM
2008 Oracle Corporation
29
File #1
File #2
Online Provisioning and Capacity Management
New Cell
-
Protection from Hardware Failure
ASM implements mirroring at the
allocation unit level
Primary and mirror copy of allocation
units are placed on separate storage
cells
Automatically remirrors across all
DB
File #1
DB
File #2
ASM
2008 Oracle Corporation
30
Automatically remirrors across all
remaining cells when a disk or array fails
Failure of disk or array is transparent to
database
Designed to Tolerate Failures of Cells and Disks
-
Protection from Brownout
ASM temporarily freezes I/O to
unresponsive cells
Redirect read IOs to mirror copy
Track write IOs that are m
issed
Writes replayed when cell begins
responding
Fast mirror resync
No need to remirror all the unchanged
DB
File #1
DB
File #2
ASM
2008 Oracle Corporation
31
No need to remirror all the unchanged
data
Benefits
Fast recovery from transient failures
e.g Cell crash or temporary hang
Can be used for planned m
aintenance
Cell software or component upgrade
Freeze I/O
to Cell
Designed to Gracefully Tolerate Cell Brownouts
-
Data Protection Solutions
All single points of failure elim
inated by the Exadata Storage
architecture
Hardware Assisted Resilient Data (HARD) built in to Exadata Storage
Prevent data corruption before it happens
Data Guard provides disaster protection and data corruption
protection
Automatically m
aintained second copy of database
2008 Oracle Corporation Proprietary and Confidential
32
Flashback provides human error protection
Snapshot-like capabilities to rewind database to before error
Recovery Manager (RMAN) provide backup to disk
Archiving and corruption protection
Can be used with Oracle Secure Backup (OSB) or third party tape
backup software
These work just as they do for traditional non-Exadata storage
Users and database administrator use familiar tools
-
Exadata Storage Management & Administration
Enterprise Manager
Manage & administer Database and ASM
Exadata Storage Plug-in
Enterprise M
anager Grid Control Plug-in to monitor &
manage Exadata Storage Cells
Comprehensive CLI
2008 Oracle Corporation Proprietary and Confidential
33
Comprehensive CLI
Local Exadata Storage cell management
Distributed shell utility to execute CLI across m
ultiple
cells
Integrated Lights Out Management -ILOM
Remote management and administration of hardware
-
2008 Oracle Corporation Proprietary and Confidential
34