24/02/2024

CAPACITY PLANNING &

DISASTER MANAGEMENT

CAPACIT Y PLANNING IN THE CLOUD

 Cloud computing of fer s incredible flexibility and fantastic

options compared to traditional computing or storage
methods. Many cloud ser vices operate with pay -as-you-go
models that have no boundaries on capacity or usage. The
question becomes; is capacity planning necessar y in the
cloud?
 One of the biggest advantages cloud computing of fers is the
elastic usage of resources. The cloud requires less hardware
than traditional computing structures which can mean greater
flexibility and a lower upfront cost. Even better, it can be easy
and quick to purchase additional resources. However,
elasticity is only good if your implementation strategy is good.

1
 24/02/2024

CAPACIT Y PLANNING IN THE CLOUD

 One of the most impor tant reasons to utilize proper planning

strategies is cost. Cloud models can seem appealing because
they of fer lower cost alternatives to previous structures.
However, there is a dangerous risk of wasting time and
money. Moreover, the allure of limitless resources may cause
companies to needlessly overspend.
 Capacity planning seeks to match demand to available
resources. Capacity planning examines what systems are in
place, measures their per formance, and determines patterns
in usage that enables the planner to predict demand.
Resources are provisioned and allocated to meet demand.

CAPACIT Y PLANNING IN THE CLOUD

 Although capacity planning measures per formance and in

some cases adds to the exper tise needed to improve or
optimize per formance, the goal of capacity planning is to
accommodate the workload and not to improve efficiency.
Per formance tuning and optimization is not a primar y goal of
capacity planners.
 To successfully adjust a system's capacity, you need to first
understand the workload that is being satisfied and
characterize that workload. A system uses resources to satisfy
cloud computing demands that include processor, memor y,
storage, and network capacity. Each of these resources has a
utilization rate, and one or more of these resources reaches a
ceiling that limits per formance when demand increases.

2
 24/02/2024

CAPACIT Y PLANNING IN THE CLOUD

 It is the goal of a capacity planner to identify the critical

resource that has this resource ceiling and add more
resources to move the bottleneck to higher levels of demand.
 Scaling a system can be done by scaling up ver tically to more
power ful systems or by scaling out horizontally to more but
less power ful systems.
 The ultimate goals of capacity planning are optimization of IT
resources and minimizing infrastructure cost. An ideal
capacity plan strikes the right balance between infrastructure
cost and the availability/per formance of the application.
 Cloud elasticity allows one to cater to the unknown scale
requirements of an application. Therefore the cloud paradigm
switches from more static and costly capacity planning
process, to a more dynamic and ongoing capacity
management process.

CLOUD CAPACIT Y MODEL

 The traditional capacity -planning process is typically achieved in

four steps:

 Step 1 . Create a capacity model that defines the key resources

and units of growth.
 Step 2. Create a baseline to understand how your ser ver, storage,
and network infrastructure are used by capturing secondary
indicators such as CPU load or global network traf fic.
 Step 3. Evaluate changes from new applications that are going to
run on the infrastructure and the impact of demand “bursting”
because of increased activity in given services.
 Step 4. Analyze the data from the previous steps to forecast
future infrastructure requirements and decide how to satisfy
these requirements.

3
 24/02/2024

CHALLENGES IN THE APPROACH

 The main challenge with this approach is that it is ver y

focused on the technology silos, not the platform as a whole.
For example:
 It views all workloads as equal and doesn’t focus on the
relative value of a mission critical workload over a noncore
application.
 It is not optimized to cope with variable demand.
 It typically will not factor in emerging data center costs such
as power consumption.
 A new capacity -planning process should attempt to provide
data that can be used to deliver more business value to the
provider and optimize the existing infrastructure and growth.

CLOUD CAPACIT Y MODEL

Cloud C.apaclty Model

I component>
_.
---' ~:-- _.
I service/Domai'> . ..-

~
I I!l§IIgur!:t" I I !!I§"""FI!!! I I I
1"'~IlliW! I I~Tjm'l
>
Busines

dl;~
Cloud .Pl.atform Forecast 8

4
 24/02/2024

CAPACIT Y MANAGEMENT

 A mature capacity management process will utilize a

combination of scale-units and resource usage monitoring
mechanisms . Scale-units consist of a set of resources such as
vir tual machines, databases, storage volumes etc. They share
a common configuration and hardware baseline and don’t
share a single point of failure with the other resources.
 Therefore cloud usage monitoring becomes a critical
component in an ongoing capacity management process.
 Without a detailed capacity management strategy, moving to
the cloud may not realize its full benefits. Applications could
outrun the resource budgeting and shoot up the IT
infrastructure expenditure than budgeted. A proper capacity
management strategy will stop you from paying for cloud
resources that are not being used efficiently.

CAPACIT Y MANAGEMENT

..·····~~~~~;~·~~~~~·~·~~~;·;~;~;;;;~~I
System
Solely
Some capacity may be proyi$iooed 10 deal with
emeroencyand matked as oIf-lim~s tor normal provisioning

Slranded
This is e:apacity no longer needed but unavwble due to
mis~on~rebon Of pcIOf operatJOnal processes
)
mnr;rs::IPM )
1d1elSpa..
This will vary~ onYlha! me currentlyactiveis

Active
This is typicaIy measuredas ~ and peak
Usable
Capacity

Demand- and Forecast-Based Capacity Management 10

5
 24/02/2024

DISASTER
MANAGEMENT

11

CLOUD DISASTER RECOVERY

 Hosting an organization's applications in the cloud is

becoming increasingly popular and is percolating down to
small businesses too. Some of the advantages of hosting
applications in the cloud are:

 Large capacity data centers

 Back-up power availability 24x7x365
 Pay -as-you-go facilities means the business pays only for the
ser vices it uses
 Real-time data back-up by the cloud ser vice provider
 Easy for organization to scale-up its cloud requirements
 There are many more advantages other than the above.
However, IT Manager s, should do due diligence on the disaster
recover y aspect of cloud ser vices.

12

6
 24/02/2024

DISASTER RECOVERY PLAN

 There are dif ferent DR approaches to develop a recover y plan

in cloud system. They are based on the nature of the system.
However in the literature, all these approaches are based on
redundancy and backup strategies. The redundancy strategy
uses separated parallel sites which have the ability to star t up
the applications after a disaster; whereas backup strategy
uses replication technology.
 The objective of disaster recover y planning is to minimize
RTO, RPO, cost, and latency by considering system constraints
such as CPU, network and storage requirements. So we can
say DR recover y planning can be considered as an
optimization problem.

13

DISASTER RECOVERY TECHNIQUES

 Hot Backup Site

▪ It is very expensive to operate. This site works with organizations that
operate real time processes. It is the duplicate of the original site.
Loss in data is very minimal as we can relocate the data and
continue our work what we are performing. It will save as a virtual
image of our current data. In a few hours hot backup site can bring
up to full production. It is priory used in the situations where disaster
happening.

 Cool Backup Site

▪ It is the least expensive to operate. It doesn’t take any backup of
data copies or it doesn’t include hardware. Lack of hardware can
start-up with a minimal cost but require more time. Everything
required to restore service to users must be procured and delivered
to the site before recovery operation is performed.

14

7
 24/02/2024

DISASTER RECOVERY TECHNIQUES

 Warm Backup Site

▪ It is already stocked with a hardware configuration on the backup
site that found in primary site. To apply warm backup site the last
data backup should be delivered to their primary sites.

Model Synchronize Recovery Backup Tolerance

Time Time Characteristics Support
Hot Seconds Minutes Physical Mirroring Very High
Warm Hours 1-24 Hours Limited Physical Moderate
Mirroring
Cold Days More than 24 Off Site backup Limited
Hours

DR Models 15

DISASTER RECOVERY CHALLENGES

 DEPENDENCY
▪ One of the disadvantages of cloud services is that customers do not
have control of the system and their data. Data backup is on
premises of service providers as well. This issue makes dependency
on CSPs for customers (such as organizations) and also loss of data
because of disaster will be a concern for customers.

 FAILURE DETECTION
▪ Failure detection time strongly affects on the system downtime, so it
is critical to detect and report a failure as soon as possible for a fast
and correct DR.

16

8
 24/02/2024

DISASTER RECOVERY CHALLENGES

 SECURIT Y
▪ DR can be created by nature or can be human-made. Cyber-terrorism
attack is one of human-made disasters which can be accomplished for
many reasons. In this case, protection and recovery of important data
will be a main goal in DR plans beside of system restoration.

 REPLICATION LATENCY
▪ DR mechanisms rely on replication technique to make backups. Current
replication techniques are classified into two categories: synchronous
and asynchronous. However, both of them have some benefits and some
flaws. Synchronized replication, guarantees very good RPO and RTO, but
it is expensive and also can affect on system performance because of
large overhead. This issue is more serious in multi-tier web applications,
because it can significantly increase Round Trip Time (RRR) between
primary and backup site. On the other hand, a backup model adopted
with asynchronous replication is cheaper and also system suffers low
overhead, but the quality of DR service will be decreased.

17

DISASTER RECOVERY
AS A SERVICE

18

9
 24/02/2024

DISASTER RECOVERY AS A SERVICE

 Disaster recover y as a ser vice is an upcoming ser vice as a

nomenclature of cloud computing. It is a low cost ser vice
when compared to traditional disaster recover y. It is flexible
in replicating physically or vir tually. It provides application
consistent recover y for some working applications like SQL
ser ver. It has pre-built options for vir tual recover y
environments including security, network connectivity and
ser ver failover when continuously replication among ser vers.
When disaster occurs we can take backup and we can run our
applications on ser vice provided by disaster recover y until we
get backup to primar y site. Disaster recover y as a ser vice to
replicate critical ser vers and data centre infrastructure in
cloud.

19

DISASTER RECOVERY AS A SERVICE

Disaster Recover y as a Ser vice

20

10
 24/02/2024

DISASTER RECOVERY AS A SERVICE

 Disaster recoveries as a ser vice is free or pay on use of fer.

When incompatibilities are occurred due to sof tware changes
then breaking of DRaaS in cloud may occur. The architecture
of DRaaS is defined by three models.
▪ From Cloud When the primary application or data is in cloud and
backup or recovery site is in private data centre.
▪ In cloud When both primary site and recovery site are in cloud.
▪ To cloud When the application is in primary data centre and backup
or recovery site is in cloud. To test the recovery processes sandboxes
are used and they test without disrupting running application. It is
only accessible to only system administrator.
 Solutions are pre-packaged ser vices that provide a standard
DR Failover to a cloud environment that you can buy on a pay -
per-use basis with var ying rates based upon your recover y
point objective (RPO) and recover y time objective(RTO).

21

DR SOLUTION CAPABILITIES

 Following must be par t of a DR solution at a minimum:

▪ The ability to automatically backup critical systems and data.
▪ The ability to quickly recover from a disaster, with minimal user
interaction.
▪ Flexible recovery options, such as restoring a single application or
the whole infrastructure.
▪ Easily understood billing structure.
▪ Backup Target Options.

22

11
 24/02/2024

TRADITIONAL DR vs DRaaS

..,.
K:::I __

-=
,

:= ~-=: ~ A
..,.
!

.,
:
-=::J __
..J
BACKUP
".. :::: ::::
••••••••
.
Backup Appliance
RESTORE
. (f) ~._/
-=::2
Run
'1"'-" .

Secoodary Hardware
-= Or Site

23

12