Embed Size (px)
Citation preview
Sql Server
For The
Experienced
Oracle DBA
Aaron Shilo
CEO@DBcs
www.dbcs.co.il
About me
• Maried + 3
• Was CTO @ johnryce israel
• Currently Owner and CEO @ DBcs ltd
• Certified as ocp and mcp
1
DBcs Ltd
• Provide Premium DataBase Services.
• Specialize in :
– Design : logical , physical
– Integration , migration , upgrade in and between
Sql Server and Oracle dbms.
– Database management and development using
latest techniques and technologies.
– Great troubleshooting skills
Among our customers
2
Agenda
• Understand the underlying architecture of SQL Server– Services
– Instance Structure
• Manage the SQL Server system, databases, and users– Database Structure
– Database Option/Parameters
– Users & Roles
– Security Concepts
– Advanced Security Overview
• Database backup & recovery– Recovery Model Options
– Backup Options
– Restore Options
– Advanced High Availability Overview
• Define and implement monitoring and tuning solutions for the SQL Server system– Resource Governor
– Profiler
– Database Tuning Advisor
Instance Architecture
3
Concept of Database and Instance
• Database Management System (DBMS) encompasses database and instance
• A Database is a set of physical files that store application and system data
• An instance consists of the hardware resources of the system (primarily memory and CPU utilization) that facilitates data access for the end-user
Database and Instance Implementations
4
Database and Instance Implementations
Database and Instance Implementations
5
Database and Instance Implementations
Database and Instance Implementations
6
Client Interaction with Database and Instance
• Key components in the interaction of the user (client) with the database environment
Application
Layer
-------------
SQL Server
Network
Interface
protocol
layer
-------------
SNAC
network
libraries
ClientPerson
ClientSystem
Database ServerEngine
TDS Endpoint
(defined by
listener.ora
may implement
TNSnames.ora)
Relational
Engine
--------------
Tabular Data
Stream (TDS)
Packet
------------
SQL Server
Network
Interface (SNI)
Layer
StorageSystem
Data Storage
Application
Layer
-------------
Oracle Net
Foundation
Layer
-------------
Oracle
Protocol
Support
Layer
ClientPerson
ClientSystem
Database Server
Listener
(defined by
listener.orga
may implement
TNSnames.ora
Oracle
Instance
--------------
Oracle Net
Foundation
Layer
------------
Oracle
Protocol
Support
Layer
StorageSystem
Data Storage
Oracle
Connections
SQL Server
Connections
Configuration of a Server
• Using Configuration Parameters to Allocate
and Manage Resources
• Oracle Configuration• Use of initialization parameters (init.ora or spfile)
• ALTER SYSTEM, ALTER DATABASE
• SQL Server Configuration• Setting server properties using SQL Server Management Studio (SSMS)
• The role of sp_configure system stored procedure
7
Memory Address Space – Oracle
• Components of Oracle in Me
Application components
RDBMS components
SYSTEM GLOBAL AREA
JAVA POOL
LARGEPOOL
STREAMSPOOL
SHAREDPOOL
REDOBUFFER
BUFFERCACHE
OracleProcess
Server Process
Server Process
Server Process
BackgroundProcess
PGA PGA PGA
PGA PGA
Automatic Shared MemoryManagement
Managed by SGA_Target
Automatically sizes
SQL Server’s Memory Pool and Oracle’s SGA
Oracle System Global Area (SGA)
Buffer Cache Shared Pool
Large
Pool
Java
Pool
Redo
BuffersFixed
SGA
Streams
Pool
SQL Server Memory Pool
Buffer Cache Procedure Cache
Connection
ContextLog Cache System
Structures
CLR Hosting
Layer
8
SQL Server Architecture (SQLOS)
• Components of SQL Server Architecture
• Non-Preemptive Scheduling
• Memory Management
Hosting Subsystems
Memory Management Non-Preemptive
Scheduling
SQLOS
Memory Nodes
Memory Clerks
Caches
Pools
Memory Objects
Scheduling Nodes
Schedulers
Tasks
Worker Threads
System Threads
Exception Handling Resource Monitoring
Memory Address Space – Comparison (Cont.)
• Memory Allocation in SQL ServerDynamic at
higher level (memory pool, etc.)
lower level (buffer cache, procedure cache, etc.)
Few configurable parameters
• Memory Address Range32-bit (x86)
Address WindowingExtensions (AWE)
/3GB
64-bit (x64 and xIA64)
2TB
9
Process and Thread-Based Architecture
• Process - memory space to protect applications
• Threads –
Execute code
Allocated time on CPU
Managed by Operating System
• Fibers are light-weight threads
reduce context switching overhead
• Oracle uses processes in Unix and threads in Windows
• Both Oracle and SQL Server use threads and optionally fibers.
• Allocate thread assignment with CPU affinity
Resource Governor = Greater Control
Resource Governor
• Performance
• Availability
• Security
Key Components
• Resource Pools
• Workload Groups
• Classification
10
Oracle Client–Database Interaction
• Oracle’s two modes of interaction are:• Dedicated Server – uses a dedicated server process per client
session• Shared Server – uses dispatchers and shared server processes
DATABASE Data Files
USERS
User
Process
User
Process
User
Process
User
Process
User
Process
User
Process
User
Process
User
Process
Client
Client
Client
Request
Queue
Database
Buffer
Cache
Response
Queue
Dispatche
r
Shared
Server
Dedicated
Server
INSTANCE
SGA1
5
6
7
2
3
4
8
SQL Server Client–Database Interaction
• SQL Server uses the shared server mode• Functionality of the dispatcher is facilitated by the networker thread
• Functionality of the shared server processes by the worker threads
Users
User
Process
User
Process
User
Process
User
Process
User
Process
User
Process
User
Process
User
Process
Client
Client
ClientDatabase
Buffer
Cache
Data Files
Relationa
l
Engine
OLE
DBStorage
Engine
Open
Data
Services
Worker
Thread
Worker
Thread
Worker
Thread
Memory
Pools
11
Oracle Background Processes
USERS
User
Process
User
Process
User
Process
User
Process
User
Process
User
Process
User
Process
User
Process
Client
Client
Client
INSTANCE
DATABASE
Automatic Storage Management
Process
Monitor
System
Monitor
Queue
Monitor
Job
Queue
Event
Monitor
Recoverer
Trace
Writer
Dispatcher
Shared
Server
Request
Queue
Response
Queue
SGA
Redo
Log
Buffer
Data FilesArchive Log
Files
Control
Files
Redo Log
Files
Database
Buffer
Cache
Dedicated
Server
Databas
e
WriterCheckpoint
Archive
r
Log
Writer
SQL Server Background Processes
Users
User
Process
User
Process
User
Process
User
Process
User
Process
User
Process
User
Process
User
Process
Client
Client
Client
Relational
Engine
OLE
DB
Storage
Engine
Data Files
Log
Files
Database
Buffer
Cache
Log
Cache
Database
Cleanup
Worker
Thread
Log Writer
Lazy
Writer
Worker
Thread
Database
Shrinking
Open
Data
Services
Memory
Pool
12
Comparing Background Processes
Oracle
Process
Oracle
Identifier
SQL Server
Equivalent
Required
Process Monitor PMON SQLOS Mandatory
System Monitor SMON Database
Cleanup/Shrinking
Mandatory
Database Writers DBWn Lazywriter Mandatory
Checkpoint Process CKPT Database Checkpoint Mandatory
Recoverer RECO MS DTC Optional
Log Writer LGWR Log Writer Mandatory
Archive Processes ARCn N/A N/A
Job Queue Processes Jnnn SQL Agent Optional
Job Queue Coordinators CJQn SQL Agent Optional
Queue Monitor Processes QMNn SQL Agent Optional
Parallel Query Slave
Processes
Pnnn Worker Threads Mandatory
Dispatcher Dnnn Networker Thread Mandatory
Shared Servers Snnn Worker Threads Mandatory
Database Architecture
13
Database Capacities
• A brief look at Oracle and SQL Server 2008 maximum capacity specifications
Database Capacities on Windows
Server 2003 (x86, x64, or IA64)
Oracle 10g SQL Server 2008
Minimum data block size 2 KB 8 KB
Maximum data block size 16 KB 8 KB
Maximum file size 64 GB 16 TB
Maximum number of files 65,536 32,767
Maximum database size 4 PB
(65,536 * 64 GB)
524 PB
Maximum control file size 20,000 data blocks N/A
Instances per server Limited by host resources 50 – Standard and
Enterprise Edition
16 or less on other
editions
Maximum database extent 4GB 64 KB
Maximum number of tablespaces 65,536 32,767 (filegroups – see
Module 2)
Maximum log file size 64 GB 2 TB
Schema and Data Structures (Objects)
• Schema – a collection of objects owned by a database user
• Schemas in SQL Server provide logical separation of objects,
similar to Oracle’s schema
Oracle SQL Server
Table Table
Index Index
View View
Synonym Synonym
Sequence Identity Columns
Procedure Stored Procedure
Function Function
Package N/A
Queue in Streams Advanced Queuing Service Broker Queue
Object Type Type
XML DB XML Schema Collection
Comparison of Core Schema and Data Structures (Objects)
14
Storage Architecture
• Database storage architecture includes physical and logical structures
• Physical structures are data files, log files, and operating system blocks
• Logical structures are subdivisions of data files used to manage storage space
Data File Data File Data File Data File Data File Data File
Temporary Tablespace Groups
Tablespace
ExtentExtent
FilegroupTablespace
Segment Segment
ExtentExtent
Filegroup
Heap/Index Heap/Index
ExtentExtent Extent
BlocksBlocksBlocks Blocks Pages Pages Pages
Data Blocks, Extents, and Segments
Structure Oracle SQL Server 2008
Smallest unit of logical storage
Block Page
Block size Variable 8 KB fixed
Storage allocationPerformed in multiple blocks; are ‘extents’
Performed in multiple pages; are ‘extents’
Extent size Variable 64 KB fixed
SegmentAny logical structure that is allocated
storageNo equivalent structure
15
Storage Allocation
Fundamental difference in storage allocation between Oracle and SQL Server
Managing Extents and Pages
• In Oracle, each extent is dedicated to an allocated object. In SQL Server, the equivalent is a uniform extent.
• SQL Server uses mixed extents: pages are allocated to objects with less than 8 blocks of data
• Similar to the Oracle bitmap functionality used to manage free space and extent allocation, SQL Server uses the Global Allocation Map (GAM) and Shared Global Allocation Map (SGAM)
• Oracle keeps track of extents using extent allocation maps
16
Managing Extents and Pages (Continued)
Current Use of ExtentGAM Bit
Setting
SGAM Bit
Setting
Free, not being used 1 0
Uniform extent, or full
mixed extent0 0
Mixed extent with free
pages0 1
File Header Extent
Extents in
SQL Server
GAMs and SGAMs
Tablespaces and Datafiles
• Oracle and SQL Server store data in datafiles
• The largest logical storage structure in Oracle is a tablespace
• The largest logical storage structure in SQL Server is a filegroup
• Tablespaces/filegroups are used to group application objects
• Tablespaces/filegroups optimize administration of datafiles
17
SQL Server Data Files
Three file types supported by SQL Server:
• Primary Data Files
• Secondary Data Files
• Log Data Files
Mapping of Tablespaces and Filegroups
System Tablespace
SysAux Tablespace
Temporary Tablespace
BigFile Tablespace
User Data Tablespace
User Index Tablespace
Undo Tablespace
Redo Log Files
Master DB
Resource DB
TempDB
Model DB
MSDB
User DB
Data FG
Index FG
Log File(s)
Oracle Database Instance SQL Server Instance
Data file
Data fileData file
Data fileData file
Tablespace Group
Data fileData file
Data fileData file
Data fileData file
Data fileData file
Data fileData file
Redo LogRedo Log
Redo Log
Log fileData file
Log fileData file
Log fileData file
Log fileData file
User Database
Data fileData file
Data fileData file
Log file
Log fileData file
18
Logging Model
• Oracle uses online redo logs to record changes made to the database by transactions and undo segments to capture the ‘before image’ of data
• SQL Server implements both of these functions using transaction logs. Each transaction record contains the undo and redo image of the transaction.
Data Dictionary
• In Oracle the data dictionary is stored under the SYS schema in the
SYSTEM tablespace
• In SQL Server the data dictionary consists of:
– Catalog View—the best way to access system metadata
– Backward Compatibility Views—All system tables from previous
releases are provided as backward compatibility views
– Dynamic Management Views—to view the current state of the SQL
Server system. Provide real-time snapshots of internal memory
structures indicating the server state.
– INFORMATION_SCHEMA views—SQL-99 method to view system
metadata
• The SQL Server Resource database contains the metadata for system
stored procedures
19
Basic Administration
Installing SQL Server 2008
Stages in SQL Server 2008 installation• Planning—software edition, file locations, naming
conventions, installation options (SQL Server Versions, OFA recommendations: file locations)
• Resource procurement—hardware, operating system, utilities
• Installing software and components
• Network configuration—choice of protocols, load balancing, etc., using SQL Server Configuration Manager
• Client connectivity and configuration—choice of protocols, server details using SQL Server Configuration Manager
20
Database System
• Creating instances – Initial and subsequent instances can be named; one default, unnamed instance
• Configuration – Database System configuration parameters can be set using sp_configure or Management Studio
• Configuration parameters can be basic or advanced – Certain parameters are self-configuring and can also be manually set
SQL Server Databases
• Database name — constrained by the same
rules as identifiers
• Database/Schema owner — user that owns
database and objects
21
Database Configuration
• Database configuration option set using ALTER DATABASE and sp_dboptions can be classified as:
• Auto options – auto_close, auto_shrink, …
• Cursor options – address cursor restrictions and scope
• Recovery options – recovery full | bulk_logged | simple
• SQL options – ansi_padding, …
• State options – offline | online, single user, …
Database Maintenance
• Information on the database can be obtained using sp_helpdb
• Database maintenance operations in SQL Server and their Oracle
equivalent are:
• Shrinking database – equivalent to shrinking all the files in a
tablespace
• Changing default filegroup – same as changing default tablespace but
set for the database rather than the user login
• Attaching & detaching databases – similar to transporting tablespaces
• Changing database name – similar in function to Oracle’s tablespace
renaming feature
• Changing database owner – not found in Oracle
22
Data Dictionary
• SQL Server term for data dictionary is system catalog
• System catalog is decentralized with instance system objects in the Master and Resource databases and database-specific objects in individual databases
• INFORMATION_SCHEMA views provide an interface to schema related system tables
• SQL Server provides system catalog views to obtain metadata – equivalent to Oracle’s DBA_ / ALL_ / USER_ views
System Stored Procedures
System Stored Procedure categories include:
• Active Directory • Catalog • Cursor• Database Engine• Database Mail
and SQL Mail• Distributed Queries• Full-Text Search• Log Shipping• Automation
Notification Services
Replication
Security
SQL Server Profiler
SQL Server Agent
Web Task
XML
General Extended
23
Data Objects
Schema Objects – Table
• Tables are the primary objects that store data in rows and columns
• Each DBMS offers different types of tables to support different access needs with optimal performance and extended functionality as the goals
Oracle SQL Server
Heap-Organized Table
Heap
Clustered Table -
Partitioned Table Partitioned Table
Nested Table XML datatype
Temporary Table Temporary Table
External Table Linked Server
Object Table Table Types
Index-organized Table
Clustered Index
Table Types in Oracle and SQL Server
24
Putting Data in the Table
• Heap is the default form of data organization in a table, where data is stored in an unordered manner
• Allocation units can extend beyond themselves
• Oracle’s index organized tables store data in a B-tree structure in primary key order
• Oracle and SQL Server offer local and global temporary tables to store transient application data
• Compression is necessary
SQL Server Table Structures
Clustered Index Structure in SQL ServerSysindexes
Root
node
Intermediat
e level
Leaf node /
data pages
A-Z
W-ZA-D E-H …
…
Debr
a
Andy Charlie
25
Indexes
Index Structures in Oracle and SQL Server
Index scheme Oracle SQL Server
B-tree Unique Yes Yes
B-tree Non-unique Yes Yes
B-tree Composite Yes (32 cols) Yes (16 cols)
B-tree Ascending Yes Yes
B-tree Descending Yes Yes
B-tree Cluster Yes Yes
B-tree Reverse key Yes No
B-tree Key compressed Yes No
B-tree Function-based Yes No
B-tree Index organized table Yes Yes (clustered)
B-tree Partitioned Yes No
Bitmap Yes (30 cols) No1
Bitmap Join Yes No1
Constraints
• Constraints are data integrity rules defined on columns of a table to enforce certain business rules
Oracle SQL Server
NOT NULL NOT NULL
UNIQUE UNIQUE
PRIMARY KEY PRIMARY KEY
FOREIGN KEY FOREIGN KEY
DEFAULT DEFAULT
CHECK CHECK
Constraints in Oracle and
SQL Server
26
Triggers
Trigger Types Compared
Feature Oracle SQL Server
DML – INSERT Yes Yes
DML – UPDATE Column/Row Row
DML – DELETE Yes Yes
Timing – BEFORE Yes Yes (INSTEAD OF)
Timing – AFTER Yes Yes
Level Row/Statement Row
Views – INSTEAD OF Yes Yes
Multiple triggers per action Yes Yes (first/last specified)
DDL Triggers No Yes
Single trigger for multiple actions Yes Yes
Views
• Both Oracle and SQL Server offer views based on simple queries involving a single table and complex queries based on multiple tables
• Indexed views in SQL Server is the only type of view which actually stores data
• Oracle and SQL Server offer updatable views with INSTEAD OF triggers and WITH CHECK OPTION constraint
View Type Oracle SQL Server
Simple views Yes Yes
Join views Yes Yes
Partitioned views Yes Yes
Updatable views Yes Yes
Inline views Yes Yes
Object views Yes No
Indexed views No Yes
View Types in Oracle and SQL Server
27
Number Generation and New Names
• CREATE SEQUENCE SomeNewNumbers
• START WITH 0 INCREMENT BY 1
ORACLE: Sequence ( 0, 1, 2, 3, 4, ……….
• CREATE TABLE ASchema.ATable
• ( Id int IDENTITY(0,1),
• Information nvarchar(12) )
SQL Server: Identity (0, 1, 2, 3, 4, ………..
• CREATE TABLE dbo.Globally_Unique_Data
• (guid uniqueidentifier CONSTRAINT Guid_Default DEFAULT NEWSEQUENTIALID() ROWGUIDCOL,
• Employee_Name varchar(60)
• CONSTRAINT Guid_PK PRIMARY KEY (guid) );
SQL Server: Global Unique Identifiers( B2794358-DA6A-DD11-829F-
001E685F0BC5…
• CREATE SYNONYM MyProductFOR TheSchema.TheProduct;
Either: Synonyms
Character-Based Data Types
Character-based data types in Oracle and SQL Server
Oracle SQL ServerData Type Size (bytes) Data Type Number of
CharsSize in bytes
Char 1 to 2000 Char 1 to 8000 1 to 8000 fixed
NChar 1 to 2000 (fewer chars) NChar 1 to 4000 2 to 8000 fixed
Varchar 1 to 4000 Varchar
Varchar(max)
1 to 8000
1 to 231-1
0 to 8000
0 to 2 GB
NVarchar 1 to 4000 (fewer chars) NVarchar
NVarchar(max)
1 to 8000
1 to 230-1
0 to 4000
0 to 2 GB
Varchar2 1 to 4000
NVarchar2 1 to 4000 (fewer chars)
LONG 1 to 231 Text, Varchar(max)
1 to 231-1 0 to 2 GB
CLOB 1 to 232 Text, Varchar(max)
1 to 231-1 0 to 2 GB
NCLOB 1 to 232 Ntext, Nvarchar(max)
1 to 230-1 0 to 2 GB
28
Numeric Data Types
Numeric data types in Oracle and SQL ServerOracle SQL Server
Number(19,0) BigInt
Int or Number(10,0) Int
SmallInt or Number(6,0) SmallInt
Number(3,0) TinyInt
Number(p,0) Decimal(p,s)
Number(p,0) Numeric(p,s)
Float or DoublePrecision or Number(38)
Float
Number(1) Bit
Number(19,4) Money
Number(10,4) SmallMoney
Date and Time Types
Oracle SQL Server
Data Type Values Data Type Values
Date Date and time to seconds SmallDateTime Date and time to seconds
DateTimeDate and time with fractional seconds
to 1/300 or 3.33 milliseconds
Calendar01/01/1753 AD – 01/06/9999 AD
(DMY)
Timestamp (TS)Date and time with fractional seconds
(9 digits)DateTime2 (DT2)
Date and time with fractional seconds (7 digits)
Timestamp with time zone (TSTZ)
Like TS with zones
Timestamp with local time zone
(TSLTZ)
Like TS with relative zones to users
DateTimeOffset Like DT2 with time zone offsets
Calendar – Julian01/01/4712 BCE -
12/31/4712 CECalendar –Gregorian
01/01/0001 AD – 31/12/9999 AD (DMY)
Daylight Savings Time Support
YesDaylight Savings
Time SupportNo
DATE Date only – Gregorian
TIME Time only with fractional seconds(7 digits)
29
Binary Data Types
Oracle SQL Server
BLOB Image, Varbinary(max)
Raw Image, Varbinary(max)
Long Raw Image, Varbinary(max)
BFile Similar to
Varbinary(max) Filestream
BLOB/Raw(n) Binary(n)
BLOB/Raw(n) Varbinary(n), Varbinary(max)
Binary data types in Oracle and SQL Server
Database Security
30
Managing Users
• In Oracle, a username is database system wide, SQL Server uses login accounts to access the instance and user accounts for the individual databases
• Oracle usernames and SQL Server logins can be operating system authenticated or database authenticated. In addition, SQL Server logins could be authenticated by the network domain.
• In SQL Server, a user account has to be created in every database that a login needs access to and can be named differently from the login name
Managing Privileges
• Oracle and SQL Server control access and activity within the database using system and object privileges
• ALTER DATABASE and GRANT are examples of system privileges while object privileges can be SELECT, INSERT, UPDATE, DELETE.
• Oracle and SQL Server use the GRANT statement to give privileges and REVOKE statement to remove privileges. In addition SQL Server has the DENY statement to suspend privileges.
• Use the system catalog view sys.database_permissionsand function fn_builtin_permissions to list permissions on objects and statements in SQL Server
31
Managing Roles
• Oracle and SQL Server provide system roles with predefined privileges and user defined roles
• The SQL Server system defined roles are of two categories:1. Fixed server roles which are defined for the database instance
2. Fixed database roles whose scope is limited to a database
• Sysadmin fixed server role is equivalent to Oracle’s DBA role
• In Oracle, there is a single DBA role that has database instance wide privileges spanning all schemas
• In SQL Server, administrative privileges can be limited to individual databases by the use of fixed database roles
Backup and Recovery
32
Database Errors
The most common database errors include:
• Statement failure—when a program encounters a failure in the handling
of a statement
• Process failure—when a background process, user process, or server
process encounters an abnormal disconnection or process termination
• Instance failure—when an instance ceases to run due to a hardware or
software failure
• User or application error—when objects are dropped or data is deleted
accidentally
• Media failure—when reading from or writing to a database file fails
• Network failure—when a network segment fails or a phone connection
aborts
Backup Methods
Two Types: Logical and Physical Backups • Logical backups facilitate recovery at the
individual schema object level• In SQL Server, you can use SSIS, bcp, and other
data transport tools
• Physical backups are copies of physical database files. These include:• Online (or inconsistent) backups
• Offline (or consistent) backups
• Incremental backups
• Archive log backups
33
SQL Server Recovery Models
• Available recovery models include:
• Full
• Bulk Logged
• Simple
• Recovery models determine the amount of logging at the database or command level
• Recovery models provide much needed control over logging and speed up bulk operations while providing recoverability to transaction-heavy systems
Factors Affecting Backup Strategy
Decisions
• Transaction rate and rate of data change
• Availability requirements of databases
• Value and degree of mission critical nature of data
• Nature of data access
• File system storage space available for backups
• Backup media space available
• Cost of other hardware and software resources
• Performance load requirements
34
Recovering From Database Errors
• Statement failure—Oracle and SQL Server automatically recover and
return control
• Process failure—PMON in Oracle and SQLOS in SQL Server automatically
roll back the current transaction and release held resources
• Instance failure—Oracle and SQL Server recover on instance startup
• User or application error—Oracle and SQL Server offer point in time
recovery options
• Media failure—Oracle and SQL Server offer file-level recovery
• Network failure
• PMON in Oracle and SQLOS in SQL Server recover aborted transactions
• RECO in Oracle and MSDTC in SQL Server recover from failures during
two-phase commit of distributed transactions
Backup and Recovery Tools
• Oracle Recovery Manager (RMAN)—part of Oracle, used to back up and recover database files.
• SQL Server backup and restore components—in conjunction with SQL Server Agent, you can set up, schedule, automate backups and perform recovery.
• BACKUP and RESTORE commands can be used in applications, T-SQL scripts, stored procedures, and triggers.
• Backup history is maintained in the RMAN recovery catalog. The equivalent system catalog tables are in the msdb database in SQL Server.
• Consider Encryption/Compression – Both tools provide it in some approach.
35
Third-Party Backup and Recovery
Solutions
• EMC, Compaq, Hitachi, Quest, Idera, and other third party vendors offer SQL Server backup and restore tools
• Third-party backup and recover tools interface RDBMS native backup and restore components with media management. Some examples include:• BMC’s Recovery Manager
• EMC’s Datamanager
• Symantec’s NetBackup
• HP’s Data Protector
• Red Gate’s SQL Backup
• Quest’s LightSpeed
• Idera’s SQLsafe
Scalability and High Availability
36
High Availability
• High availability — ensuring database and data are accessible for the maximum amount of time
• Initialization parameters or configuration options in Oracle and SQL Server
• Logical (add, remove, rename columns) and physical (rebuild, move, change storage parameters) structure of tables
• Indexes
• SQL Server partitioning option
Clustering
• Both Oracle and SQL Server offer Active/Passive (single instance) and Active/Active (multi-instance failover clustering) options in clustering. In Oracle, it is called Failsafe and Real Application Clusters (RAC); SQL Server refers to it as Failover Clustering
• For Active/Passive clustering, Failsafe and Failover clustering are both based on Microsoft Cluster Services (MSCS) containing a single set of non-shared disks
• For Active/Active clustering, Failsafe and Failover clustering utilize a multi-node configuration where each node has its own instance connecting to shared disks. However, the database can be opened by only one node at a time
• Oracle’s RAC can also provide Active/Active clustering in a shared-everything architecture where multiple instances can simultaneously access data from the same database
37
Standby Database
• Both offer standby database which can be used for availability and disaster recovery• In Oracle, Data Guard
• In SQL Server, Database Mirroring and Log Shipping
• Standby databases are updated using archive logs in Oracle and transaction logs in SQL Server
• Shipping of logs to a standby server and their application can be controlled to provide scope for recovery from user errors in the primary
• Snapshots of mirrored database in SQL Server and logical standby database in Oracle can be used for read-only purposes such as reporting, analysis, DSS
Replication – Objects
• Replication is a specialization of the distributed database concept where
copies of an object can be kept in multiple databases
• Unlike clustering and standby databases, replication encourages sharing
and distribution of individual objects rather than entire databases
• Both Oracle and SQL Server provide multiple types of replication based on
whether the replicas are read-only or updatable
• Replicas are refreshed in full or updated using discrete transactions either
synchronously (multi-master) or asynchronously (all others)
• Oracle Streams offers replication in three forms:
• Multi-master replication — multiple peer-to-peer updatable replicas
• Materialized views — each replica has a master site with various policies on updating the replicas
• Hybrids of replication and materialized views
38
Replication – Types
• SQL Server offers the following types of replication:
• Snapshot replication — used mostly as read-only copies which are
completely refreshed periodically
• Transactional replication — kept up-to-date with immediate or
deferred updates from the master site. Copies are used mostly as
read-only and occasionally updated. Updatable subscribers
(immediate and queued), and peer-to-peer transactional replication
support updates at subscribers.
• Merge replication — similar to Oracle’s multi-master replication where
the autonomous sites are synchronized at regular intervals
• Oracle and SQL Server offer sophisticated algorithms for conflict resolution
among updatable replicas
Database Mirroring
• Software solution for increasing the availability of a specific database
• Consists of principal server, a mirror server, and optionally a witness server
• Witness server enables automatic failover
• Can operate in either Synchronous or Asynchronous mode
• Use views such as sys.database_mirroring, sys.dm_db_mirroring_connections to monitor database mirroring
39
Scalability
• Scalability is the ability to support very large databases (VLDB) and/or large volume OLTP
• Memory areas in Oracle and SQL Server are dynamically resized without restarting. The process is performed either automatically or by the DBA in Oracle, where it is automatically performed by the DBMS in SQL Server.
• Oracle and SQL Server, respectively, use parallel query and worker threads to provide interoperation and intra-operation parallelism in the execution of DML and DDL. Both optimizers are parallel-aware.
• Oracle and SQL Server scale well in the handling of user connections through the use of shared server processes
• Oracle and SQL Server offer many types and features for speedy access to large volumes of data. Examples are many types of indexes, advanced I/O techniques and caching.
• SQL Server 2008 supports NUMA and Hot Add CPU
Scalability and HA in Backup and
Recovery
• MTTR targets can be set for instance recovery by controlling the checkpoint process in Oracle using FAST_START MTTR_TARGET and in SQL Server using ‘recovery interval’ option
• The RMAN features available in SQL Server backup and restore utilities are:• Parallelism in backup and recovery operations• Incremental backups• Multiple granularities in backup and restore to reduce the
amount of recovery needed• Point in time recovery• Choice of backup media and devices• Duplicating or cloning databases
• Moving of data in logical units of tablespaces using transportable tablespaces across multiple platforms in Oracle is found in SQL Server through the attach and detach features for databases
40
Performance Tuning
Tuning Methodologies — Tuning While
Building
• The Tuning While Building methodology offers
a proactive approach to performance tuning
• These are the steps:
1. Tune the business rules
2. Tune the data design
3. Tune the application
4. Tune the instance
5. Tune the underlying platform
41
Tuning Methodologies — Performance
Tuning
• Performance tuning is an ongoing, interactive process
• Performance Improvement Method offers steps in improving performance of a system in production:
• Define the problem
• Examine the host system and the Oracle statistics
• Examine documentation to help identify the problem
• Generate a prognosis based on captured data
• Propose a course of action and implement the changes
• Evaluate the causes of the changes
• If unsuccessful, repeat previous two steps
Application Tuning
• Tune Application Design — Leverage RDBMS features
• Example: Perform calculations and summaries on the database before fetching the results so as to leverage database resources as well as to reduce network traffic
• Tune Logical Structure of Database — Leverage supporting schema objects
• Example: choose appropriate indexes
• Tune Database Operations
• Find statements that consume most resources — Use statistics available in system catalog, such as parses, CPU time, physical reads
• Tune the statements to use fewer resources — Use coding standards and tools to analyze and tune SQL
• Tune Access Paths — Aid Optimizer in finding optimal access path, creating indexes, clustered indexes, and providing optimizer hints
42
• Tune Memory Allocation
• Proper configuration and use of the caches that make
up the SGA (Oracle) and memory pool (SQL Server)
• Tune I/O and Physical Structures
• Separation of files based on access
• Distribution of I/O load across files and devices
• Tune Resource Contention
• Reduce and/or eliminate contention for locks, latches,
block (page) headers, and queues
Instance Tuning
SQL Server Performance Tuning Tools
• Profiler — records detailed information on activity on the database server, including I/O statistics, CPU statistics, index and table scans, execution plans and several other statistics to aid in tuning SQL and T-SQL
• Performance Data Collector — fully integrated tool that collects a variety of performance metrics on one or more SQL Servers and stores the data in a central datawarehouse
• Build custom reports showing trends
• Query Editor in SSMS — graphical tool used to execute and debug queries to resolve performance issues
• Database Engine Tuning Advisor — database physical design tool to optimize indexes and partitioning
• Replaces Index Tuning Wizard from previous SQL Server releases
• Performance Monitor —Windows tool that records counters at hardware, operating system, database, and application levels
• Task Manager — used for initial review of high-level counters on the system, including utilization at a process level
43
44
