Critical SQL Server Databases: Provide HA with SQL Server Failover Clustering and Cluster Shared Volumes

Webinar

Critical SQL Server Databases:

Provide HA with SQL Server Failover

Clustering and Cluster Shared Volumes

Edwin Sarmiento

http://[email protected]@EdwinMSarmientohttp://ca.linkedin.com/in/EdwinMSarmiento

Microsoft MVP/Microsoft Certified Master: SQL Server

high

availability

high

availability

characteristic of a system that is

fu l l y operat ional

for a long period of time

high

availability

is NOT the same as

UPTIME

Just because

something is

UPdoesn’t mean it’s

available

UP

available

Recovery Time Objective

(RTO)

high

availability

When is my application coming back?

downt imehigh ZERO

(one “nine”)

% availability

90

36.5 daysper year

72 hoursper month

16.8 hoursper week

dow

ntim

e

(two “nines”)

% availability

99

3.65 daysper year

7.2 hoursper month

1.68 hoursper week

dow

ntim

e

(three “nines”)

% availability

99.9

8.76 hoursper year

43.8 minutesper month

10.1 minutesper week

dow

ntim

e

(four “nines”)

% availability

99.99

52.56 minutesper year

4.32 minutesper month

1.01 minutesper week

dow

ntim

e

(five “nines”)

% availability

99.999


25.9 secondsper month

6.05 secondsper week

dow

ntim

e

(five “nines”)

% availability

99.999


25.9 secondsper month

6.05 secondsper week

dow

ntim

e

How long does it take to restart

SQL Server

0

10000

20000

30000

40000

50000

60000

70000

1 2 3 4 5 6 7 8 9 10 11

Co

st

Time

Availability/RTO versus Total Cost of Ownership

downt imehigh ZERO

features

high

availability

https://msdn.microsoft.com/en-us/library/cc645993.aspx#High_availability

ALWAYSON AVAILABILITY

GROUPS

introduced in SQL Server 2012


GROUPS

*DATABASE (group) protection

can either be an HA and/or DR solution


GROUPS

*DATABASE (group) protection

combination of

DATABASE MIRRORING &

FAILOVER CLUSTERING


GROUPS

SQL Server

log data

SQL Server

log data

Secondary ReplicaPrimary Replica

SQL Server

log data

Secondary Replica

*maximum of EIGHT replicas

Windows Server Failover Cluster

Database Mirroring

challenge

Availability Group

SQL Server

Enterprise Edition

requires

FAILOVER CLUSTERING

*INSTANCE protection

FAILOVER CLUSTERING

can either be an HA and/or DR solution

log data

failover clustering

how it works


SQL Server

2-node clustert r a d i t i o na l

failover clustering

how it works


SQL Server

Application


log data


shared disks containing an NTFS

or ReFS volume that is made

accessible for read and write

operations by all nodes of WSFC

cluster shared

volumes *

we will only focus on Windows Server 2012 +

SQL Server


failoverlet’s look at

in the context of the

storage

SQL Server


the storage is owned by

I/O workload only goes thru that single path to the

storage

only one cluster node

at a time

SQL Server


SQL Server

AD/DNS

during failover, the storage gets

“logically” dismounted from the

failed node and remounted on the

new active/primary node


SQL Server

AD/DNS

the amount of time it takes to

bring the storage online

depends on how long it takes to

mount the storage to the new

active/primary node

SQL Server


SQL Server with CSV, there are now

multiple paths to the storage

I/O workload can be routed thru different paths to the

storage


SQL Server with CSV, no need to remount

storage on new active/primary

node after failover this reduces downtime, increases resiliency and

reliability

SQL Server


with CSV, no more

23 drive letter limitations

this means you no longer have to memorize the

alphabet in terms of failover cluster drives

SQL ServerSQL Server


with CSV, the concept of

*this means that SQL Server failover clustered

instances

no longer has a direct dependencyon the shared drive to come online

exclusive ownership

of storageis redefined

SQL ServerSQL Server

failover clustering

how it works



Application

*can be implemented with

SQL Server

Standard Edition

log data

SQL Server

failover clustering

how it works



Application

*can be implemented with

SQL Server

Standard Edition

log data log data

HA

SQL Server

What does StarWind Virtual SAN change?

StarWind Virtual SAN turns local storage into fault tolerant CSV for the local SQL cluster

•Deployment price is reduced ~3 times

•Maximized performance because of short IO path

•Disk, RAM, and Flash FT without Enterprise SQL licensing

Benefits compared to native clustering

demo

Configure SQL Server 2014

Failover Clustered Instances to

use Cluster Shared Volumes

Edwin Sarmiento

http://[email protected]@EdwinMSarmientohttp://ca.linkedin.com/in/EdwinMSarmiento

Microsoft MVP/Microsoft Certified Master: SQL Server

Critical SQL Server Databases: Provide HA with SQL Server Failover Clustering and Cluster Shared Volumes

Technology

sql server failover

storage sql server

downtime high

log data failover clustering

week downtime

node cluster t r

minutesper year

database group protection