Virtual Systems and Services

Shakeel Ahmad

Lab 10
Contents

Examine the need for increased

availability in today’s information-driven Understand the availability
options and strategies in a virtual
world
environment

Examine availability in a single .

virtual machine, in a single
virtualization host, and across
multiple virtualization hosts
 Availability
Virtualization allows less expensive and more flexible options to protect an entire data center from
an interruption due to a large-scale disaster

Increasing availability
▶ Protecting a virtual machine
▶ Protecting multiple virtual machines

▶ Protecting data centers

3
 Improving Availability
Availability (%) Downtime per Year
99 3.65 days
99.9 8.8 hours
99.99 53 minutes
99.999 (”five nines”) 5.3 minutes

• Greater degrees of availability require significant costs, often forcing companies to choose
between them
• Downtime includes planned and unplanned downtime
• You cannot choose when unplanned downtime occurs

4
 Traditional Solutions
• RAID
o Improves storage availability and performance

• Storage Multipathing
o Provides multiple paths for storage traffic
o Works well in virtualization

• NIC Teaming (aka link aggregation)

o Multiple network cards linked together for more bandwidth and alternate paths in the event
of a NIC failure
o Works well in virtual environments

• Clustering solutions like Microsoft Cluster Services or Oracle RAC

o Can be implemented in VMs
o Can be implemented across VMs and physical servers

5
 Virtual Machine Availability

• Enhanced HA
o Virtual Machine Tools can automatically reboot a failed or frozen virtual machine without intervention

• Application HA
o Tools that can restart a failed application even though the operating system is still running

• Storage and Network I/O

o QOS features that ensure critical application workloads receive adequate bandwidth in times of
contention to prevent application interruptions

• Other best practices still apply

o Employ proper security and antivirus protection to guard against deliberate attempts to interrupt services

6
 High Availability

• Clustering (two or more physical servers linked together via network and
shared storage resources)
• When one physical server fails, all the VMs on that server are automatically
rebooted somewhere else in the cluster
o Need to configure headroom during planning stages
o Part of the infrastructure, more cost effective than providing same service in a physical environment
o Still means crashed virtual machines

7
 Fault Tolerance

• Able to withstand virtualization host failures without application interruption

• Second virtual machine runs in lockstep with primary
o Upon host failure secondary becomes the primary
o New secondary created in the cluster
• Have to plan resources accordingly

8
 Live Migration (vMotion)

• The ability to migrate a running VM from one virtualization host

to another without interrupting the user application
• Maintenance mode
o Evacuates the guests from a host
o Work can be done on the physical host, or even replace the host, without disrupting the
guests or their applications
o Planned downtime can occur at any time with no disruption of service

9
 Storage Migration

• Like live migration, but in this case the files that compose a virtual
machine are moved from one place to another
• Can be used to alleviate disk contention (hot spots) without interrupting
the guest
o Can be automated to resolve performance issues
• Can be used to move data to newer storage
10
 Protecting Datacenters

• Leveraging replication solutions can provide current copies of a production site

ready to be powered on
o Still a recovery
• Virtualization eases traditional challenges (network, scaling) with disaster recover
(DR) solutions by providing a standard infrastructure
• Future solutions will provide protection for datacenters as an entity and offer
movement instead of recovery
o Bandwidth dependent
11
 Summary
• Today’s service expectations have made high availability a crucial
part of server implementations
• Traditional availability solutions and strategies can be deployed in a
virtual environment
• New features and solutions can provide better overall uptime at a
lower cost in a virtual environment than in a physical environment
• Availability capabilities cover single virtual machines, single
virtualization hosts, a cluster of hosts, or entire datacenters, to
deliver the level of service new cloud applications are demanding

12
•
Exercise
Your company’s data center has suffered a recent power outage, and
corporate applications were unavailable for two days. You have been asked to
craft a strategy to quickly continue operations in the event of another outage.
What type of availability (HA/DR/FT) would you recommend and why?
• You are asked to provide availability for an application that has been deemed
to be critical to the daily operations. Any downtime will potentially cost the
company hundreds of thousands of dollars per hour. What type of availability
(HA/DR/FT) would you recommend and why?

• How would you convince an owner of a business-critical application to move

to a virtual environment? Her major concern is sharing a server with other
virtual machines that might impact availability.