Introduction to High Availability with SQL Server

Introduction to High
Availability with SQL
Server
10/14/2014 www.linchpinpeople.com 1

John Sterrett
Linchpin People Group Principle
@johnsterrett
John.Sterrett@linchpinpeople.com
Johnsterrett.com
linked.com/in/johnsterrett
google.com/+johnsterrett
Leader Austin SQL
Server User Group
Leader Virtual Chapter
“High Availability and
Disaster Recovery”
PASS Regional Mentor
Specialties / Focus Areas / Passions:
• Performance Tuning &
Troubleshooting
• Virtualization
• Infrastructure
• Architecture
• High Availability
• Disaster Recovery
• Health Monitoring
• Capacity Management
• License Efficiency
www.linchpinpeople.com 2

You know us individually, but do
you know Linchpin People®?
Linchpin People® are database coaches and wellness experts for the
Microsoft SQL Server Ecosystem. Our team includes more than 15 of the
world’s top SQL Server experts, and we have presented over 175 combined
sessions at SQL Saturdays, PASS Summit, and other events in the past year.
Do you have specific questions about your database environment? Check
out www.LinchpinPeople.com and take a look at the range of services we
provide, or contact one of us today to learn how we can help improve your
business.
www.linchpinpeople.com 3

Agenda
• HA != DR
• Log Shipping
• Database Mirroring
• Transactional Replication
• AlwaysOn Failover Cluster Instance
• AlwaysOn Availability Groups
• Wrap Up / Question and Answer

Introduction
Definitions
• What do we mean when we say a system is
Highly Available?
• What does it mean to have a Disaster
Recovery system?

High Availability
What is High Availability
• High availability is a system design approach and
associated service implementation that ensures a
prearranged level of operational performance will
be met during a contractual measurement period. –
Wikipedia
• A System Design that allows for minimal downtime.
• Protection from OS or hardware failure

Examples of Disasters
Photo https://www.flickr.com/photos/ross/39556863/

Most Common Disaster

Disaster Recovery
What is Disaster Recovery
• Risk Management
• Recovery Point Objective - RPO
• Recovery Time Objective - RTO
• Disaster Recovery Plan - DRP
• Backups are not enough

Options
SQL Server Options for HA/DR
• Log Shipping
• Database Mirroring
• Replication
• Failover Cluster Instances
• AlwaysOn Availability Groups

Intro to Log Shipping
1. Transactional Log
Backup
2. Copy Transactional log backups
3. Restore Transactional log
backups
http://www.mssqltips.com/sqlservertip/2073/migrating-a-vldb-in-sql-server-with-log-shipping/

Log Shipping
• Transaction Log backups occur on the primary
• Process ships logs to secondary server(s)
• Data can be read on the secondary except
during transaction log restore
• Can have a Log Ship monitor server

Monitoring Log Shipping
• sp_help_log_shipping_monitor

Log Shipping Pros/Cons
• Works with Standard Edition
• Multiple target support
• Readable secondary copies
• Can delay restores
• Dependent on backup of
primary
• Manual failover process
• Reasonably high complexity

Database Mirroring - Modes
• High Safety (Synchronous) with Automatic Failover
• High Safety without Automatic Failover
• High Performance (Asynchronous)
Primary Server
Secondary Server
Witness
Instance
Mirror
DB
Mirror
DB

Synchronous Mode…
Mode
Principal • Transaction (Insert/Update/Delete) inserted into
Transactional Log
• Transaction read from the transactional log
• Transaction shipped to mirror
Mirror • Transaction written to mirrors transactional log
• Acknowledgement sent to principal
• Transaction committed
Principal • Acknowledgement received
• Transaction committed.

Asynchronous Mode…
(Enterprise)
Mode
Principal • Transaction (Insert/Update/Delete) inserted into
Transactional Log
• Transaction read from the transactional log
• Transaction shipped to mirror
Mirror • Transaction written to mirrors transactional log
• Acknowledgement sent to principal
Principal • Acknowledgement received

Database Mirroring
• Deprecated in SQL 2012
• Database transactions are compressed and
shipped to a secondary (2008+)
• Transfer may be sync or async (Enterprise)
• Optional witness server can facilitate failover

Warning.. Warning.. Warning..
http://technet.microsoft.com/en-us/library/ms366349.aspx

Pros and Cons
• Pros
• Witness can be any edition of SQL Server
• Page level corruption fixes
• Doesn’t require Active Directory
• Database level High Availability
• High Safety only requires Standard Edition
• Cons
• You cannot mirror system databases
• Automatic Failover requires a witness
• Database must be in full recovery
• High Performance requires Enterprise Edition
• Must create and sync Instance Objects (Jobs, Logins, etc..)

Replication
• Numerous topologies and options involved
with replication
• Snapshot Replication
• Transactional Replication
• Merge Replication
• Peer to Peer Replication

How Transactional Replication
Works
http://technet.microsoft.com/en-us/library/ms151706(v=sql.105).aspx

Replication Pros/Cons
• Replicate to multiple servers
• Replicate subset of data
• Can be done with Simple Recovery Model
• Standard Edition for transactional
• Manual failover process
• Unknown RPO
• Known to be fragile
• Re-sync can be cumbersome
• Complex

Failover Cluster Instances
Windows Failover Cluster
SQL Instance
Node 1 Node 2
Cluster Virtual Name

How Windows - SQL Clustering works
• Physical Nodes
• Windows cluster
• Windows cluster group
• MSDTC
• SQL cluster
• Other clustered application
• End User Access
27

Failover Cluster Instance (FCI)
• Pre Windows 2012 – Shared Storage
• Windows Cluster (Windows 2012 Standard Ed)
• Two Node Limit on SQL 2012 Standard
• Quorum

How Window – SQL Clustering Works
– Failover Cluster Manager
Things that you should see in
Service and Application/ Role
• Current Status
• Current Owner
• Resources
• Client Access Point
• Virtual Network Name
• IP address assigned with the name
• Storage allocated to the application
• Services / Applications or Predefined
Resources
• Others - optional
• File shares
• Printers and so on
29

Quorum

Failover Cluster Pros/Cons
• Multiple Servers (HA)
• Allows for entire instance
protection
• Automatic failover
• Transparent connections
• Complex setup
• Idle hardware
• Some cases storage single
point of failure

Introduction to Availability Groups
Instance
1
Instance
2
AG (P) AG (S)
Node 1 Node 2
Data Center 1 Data Center 2
Listner Name (VCO)
Windows Failover Cluster

Database Mirroring Problems…
• Cannot group databases
• Only get one mirror database
• No readable copies
• Requires witness (extra SQL instance) for automatic
failover

Availability Group Benefits..
• No shared storage
• Listener (Virtual Computer Object for connectivity)
• Multiple mirrors replicas
• Readable copies

AlwaysOn Availability Groups
• SQL Server Enterprise Edition
• Windows Cluster
• Async and Sync modes
• Databases failover as a group
• No shared storage needed
• Readable replicas
• Supports up to 8 replicas

AG Configurations

AG Dashboard

AlwaysOn Pros/Cons
• No shared storage
• Readable secondaries
• SSMS administration
• Configuration is easy
• Grouping DB failover helps
for complex applications
setups
• Enterprise Only - Large
setups can be
expensive
• Newer technology
• App code changes may
be needed

Questions?

Introduction to High Availability with SQL Server

More Related Content

What's hot (20)

Similar to Introduction to High Availability with SQL Server (20)

More from John Sterrett (15)

Recently uploaded (20)

Introduction to High Availability with SQL Server