You are on page 1of 22

Navigating the options for Data

Redundancy
Wendy Pastrick
SQL Server MVP

BI Consultant, BlueGranite, Chicago


About Wendy

DBA since 2000


Professional Association for SQL Server
member and volunteer
Chicago Suburban SSUG Chapter
AGENDA

Defining High Availability


A Brief Overview
Replication
Log Shipping
Mirroring
Clustering
How to Choose Which to Use?
Pros and Cons
Compare and Contrast
High Availability

Different than Disaster Recovery


Is full failover necessary?
How much downtime is acceptable?
Are personnel available to assist in a failover?
What hardware resources can be utilized?
What is the physical distance between servers?
More options than what is covered in this
presentation!
Replication

Transactional, Peer-to-Peer and Merge


options
Publishers and Subscribers
Distribution database
All databases available for use
Most flexibility with replication option
Can be complex and monitoring required
Transactional Replication

Publisher with Distribution Db

Subscriber Subscriber Subscriber


Peer-to-Peer

Node 1

Node 2 Node 3
Merge Replication

Subscriber 1

Publisher

Subscriber 2
Replication Use Case

Distributed Subscribers
Reporting Needs
Log Shipping

Transfer and application of Log File Backups


Primary to Secondary
SQL Agent Jobs
Fileshares required
Option to have Secondary database in
Standby mode for ReadOnly access
Log Shipping Topology

Backup Jobs: Database and


Logs

Secondary
Copy Job: Backup and Log
Files

Primary Employee Sales DB


DB

Employee Restore Jobs: Database and


DB Logs

Sales DB

Monitor
Log Shipping Use Case

Read Only Scenario


Built in Delay
possible security benefit
possible data retrieval
Mirroring

Transactional Application from Logs


Principal to Mirror
Utilizes End Points
Synchronous
Asynchronous
Mirroring Topology

Transactions

On Commit: Applied at Principal

Principal Mirror

Synchronous Witness
vs
Asynchronous
Mirroring Use Case

Geographically dispersed cluster


Poor mans cluster
Clustering

AlwaysOn
Availability Groups
Database Level
Windows Server Failover Cluster
Instance Level
Availability Groups

Windows Windows Windows


Cluster Cluster Cluster
Node 1 Node 2 Node 3

SQL Server SQL Server Availability


Group

DB1 DB1 DB1


Primary Secondary Secondary

DB2 DB2
Primary Secondary
Failover Clustering
Windows Windows Windows
Cluster Cluster Cluster
Node 1 Node 2 Node 3

SQL Server SQL Server SQL Server


Instance1 Instance1 Instance1

SQL Server SQL Server


Instance2 Instance2

SQL Server SQL Server


Instance3 Instance3
Pros
Replication
Multiple database support
All copies are fully functional databases
Log Shipping
Simple setup
Multiple database support
Witness for monitoring
Can use Secondary as Read-Only
Mirroring
Automatic, Immediate Failover
Least potential for data loss (in Synchronous mode)
Can use a snapshot for Read-Only of Mirror instance
Cons

Replication
Complex setup
Publications can expire
Transactions can experience
backlog/blocking
Manual Failover
Log Shipping
Asynchronous Only
Manual Failover
Mirroring
Single Database Only
Physical Distance issues
Contrast

Availability of data
Fully Recovered vs. Read-Only
Time and effort to failover
Automatic vs. Manual
Number of databases supported
One vs. Many
Summary

Any of these are viable High Availability


solutions
Make a list of questions and ensure
Management is in agreement with the
answers
Consider the available hardware and
additional resources required

You might also like