1. SQL Server
High Availability
Amit Vaid

2. Agenda How much Availability do you need?
SQL 2000 High Availability Options
Log Shipping, Replication, Clustering
Demo – SQL 2000 H.A Options
New HA options in SQL 2005
Database Mirroring
Demo – SQL 2005 Database Mirroring

3. How Much Availability do you need?
Need to ask yourself:
How long can we afford to be down?
How much data can we afford to lose?
Availability = up/(up+down)
Once you have the answers to these questions you can match the HA options to the business requirements.
the IT department generally cannot answer these questions.
When planning an HA strategy you need to look at what is already available. i.e is there a DR plan, maybe DR hardware or a DR site.
If your current H/W is struggling this can be a problem as some HA options require additional resources.
The degree to which an application or service is available when, and with the functionality, users expect

4. Barriers To Availability
Many barriers Only some are addressable by DBMS technology Be sure to consider people, planning, and procedures
Database Server Failure or Disaster
User or Application Error
Data Access Concurrency Limitations
Database Maintenance and Operations
Upgrades
Availability at Scale

5. Warm Standby Solutions Replication and Log Shipping
Database Object Level
SQL Server Replication
Database Level
Log Shipping
Both provide multiple copies and a MANUAL fail over

6. Log Shipping Minimal impact on the production server
No changes to the database are required
Transactional consistency
Supports delayed load of transaction logs
Not all SQL Server objects are automatically copied
Users must exit for next log to be applied
Failback can be time consuming

7. Demo
SQL 2000 Log Shipping

8. Replication Failover possible; a custom solution
Not limited to entire database; Can define subset of source database or tables
Copy of database is continuously accessible for read activity
Latency between source and copy can be as low as seconds
Significantly increases the on-going Database management
Database schema may need to be changed
Merge Repl – instant failback

9. SQL 2000 Transactional Replication
Demo
SQL 2000 Transactional Replication

10. Failover Clustering Microsoft Cluster Services
Hot Standby – Automatic failover
Built on Microsoft Cluster Services (MSCS)
Multiple nodes provide availability, transparent to client
Automatic detection and failover
Requires certified hardware
Supports many scenarios: Active/Active, N+1, N+I
Instance Failover – entire instance works as a unit
Single copy of instance databases
Available since SQL Server 7.0
Standby is not available for reporting, queries, etc.
May support other instances
Needs SQL 2000 Enterprise Edition
Needs Windows 2000 Enterprise Edition
Shared Nothing Arch

11. Active/Passive SQL Server Cluster
Client PCs
SQL Server
Virtual Server
Server A
Server B
Heartbeat
Cluster management
Hub
Shared
Disk Array
Hub E F G
C,D
C,D
SQL Server

12. Active/Passive SQL Server Cluster
Client PCs
SQL Server
Virtual Server
Server A
Server B
Heartbeat
Cluster management
Hub
Shared
Disk Array
Hub
SQL is shutdown and restarted on Server B. Therefore clients will be disconnected and have to reconnect. E F G
C,D
C,D
SQL Server

13. What Clustering doesn’t do:
Clustering is not a mechanism to scale
Doesn’t protect your server against site outage
Doesn’t protect your disk subsystem
Doesn’t protect against database corruption
Doesn’t protect against logical corruption
Doesn’t protect against user error
Doesn’t protect application crash
Clustering is not a method to load-balance
Still a single point of failure – The Database!

14. SQL 2000 Failover Clustering
Demo
SQL 2000 Failover Clustering

15. Log Shipping vs Clustering vs SQL Replication
So what’s the best solution….
… it depends
… On your business requirements
You can combine the SQL H.A options.
E.g. A/P Cluster with Log shipping

16. SQL Server 2005
High Availability

17. Barriers To Availability As addressed in SQL Server 2005
Database Server Failure or Disaster
Failover Clustering
Database Mirroring
Transparent Client Redirect
User or Application Error
Data Access Concurrency Limitations
Database Maintenance and Operations
Availability at Scale
User:
Good management tools, automation, and defined procedures help prevent many potential mistakes
Database Snapshot
Data Access Concurrency Limitations
Snapshot Isolation
Database Maintenance
Online Index operations
Fast Recovery
Online restore, restore a database while it remains online
Data backups don’t block log backups
Full Text catalogs backed up & restored as part of the database

18. Failover Clustering SQL Server 2005
Further refined in SQL Server 2005
More nodes
Match operating system limits
Unattended setup
Support for mounted volumes (Mount Points)
All SQL Server services participate
Database Engine, SQL Server Agent, Analysis Services, Full-Text Search, etc.

19. Database Mirroring New for SQL Server 2005
Instant Standby
Conceptually a fault-tolerant server
Database Failover
Very Fast … less than three seconds
Zero data loss
Automatic or manual failover
Automatic re-sync after failover
Automatic, transparent client redirect

20. Database Mirroring Hardware Impact to transaction throughput
Works with standard computers, storage, and networks
No shared storage components, virtually no distance limitations
Impact to transaction throughput
Zero to minimal, depending on environment / workload

21. Database Mirroring How it works
Mirror is always redoing – it remains current
Application
Witness
Commit 1 5
Principal
Mirror 2
Not two-phase commit
SQL Server
SQL Server 4 2
>2 3
>3
Log
Data
Log
Data

22. Witness and Quorum
Witness
Sole purpose of the Witness is to provide automatic failover
To survive the loss of one server you must have at least three
Prevents “split brain”
Does a lost connection mean the partner is down or is the network down?
To become the Principal, a server must talk to at least one other server

23. Witness Witness is an instance of SQL Server 2005
Single witness for multiple sessions
Consumes very little resources
Not a single point of failure
Partners can form quorum on their own

24. Safety / Performance
There is a trade-off between performance and safety
Database Mirroring has two safety levels
FULL – commit when logged on Mirror
Allows automatic failover
No data loss
OFF – commit when logged on Principal
System does its best to keep up
Prevents failover; to make mirror available
Must ‘force’ service
Or terminate Database Mirroring session

25. Transparent Client Redirect
No changes to application code
Client automatically redirected if session is dropped
Client library is aware of Principal and Mirror servers
Upon initial connect to Principal, library caches Mirror name
When client attempts to reconnect
If Principal is available, connects
If not, client library automatically redirects connection to Mirror

26. Database Mirroring Setup Steps
Ensure SQL 2005 is installed on both the principle & the mirror Server
Setup Security (Endpoints)
Prepare the mirror database
Setup & Start the database mirroring session
Optionally add a witness

27. SQL 2005 Database Mirroring
Demo
SQL 2005 Database Mirroring

28. Database States for Database Mirroring
SYNCHRONIZING
SYNCHRONIZED
SUSPENDED
PENDING_FAILOVER
DISCONNECTED
SYNCHRONIZING
The mirror is lagging behind. This is the state when Mirroring is configured.
SYNCHRONIZED
The mirror database contains the same data as the principal database. Manual and automatic failover are possible only in the SYNCHRONIZED state
SUSPENDED
The mirror copy of the database is not available. The principal database is running exposed, without sending any logs to the mirror server instance. SUSPENDED is a "permanent" state. The only ways for a session to become SUSPENDED is to be explicitly suspended or encounter redo errors
PENDING_FAILOVER
This state is found only on the principal server instance when a manual failover has been issued but not yet accepted by the mirror
DISCONNECTED
The partner loses communication with the other server instance(s) (the other partner and the witness, if one exists).

29. Automatic Failover
Automatic Failover requires the following conditions:
Database mirroring running in synchronous mode
The database must be in a synchronised state
A witness must exist
During Failover the following actions occur:
The witness & mirror server agree the primary is dead, which puts the database into a suspend state
If possible the database on the Primary server changes to a disconnected state
The mirror finishes rolling forward and records the LSN
The mirror database comes online
When the principle returns, it becomes the mirror

30. Failover Solutions At A Glance
Clustering & Mirroring both provide:
Automatic detection and failover
Manual failover
Transparent client connect
Failover Clustering
System scope
Certified hardware
Fast failover
No reporting on standby
Single copy of database
Database Mirroring
Database scope
Standard servers
Fastest failover
Limited reporting on standby
Duplicate copy of database

31. Summary How many 9’s do you NEED? SQL 2000 potential downtime ~3 mins
SQL 2005 potential downtime ~3 secs
Remember to consider process too!

32. Feel free to contact us Craig Ryan cryan@sdg.com.au Andrew Gannon
National Manager Database Services
Andrew Gannon
Business Development Manager
(03)