1. Using AHP for Decision Making
(Cloud Vs On-Premise)
By Amit Jain
2. Essence of a Scientific Tool for Decision
making
• Cloud Vs On-Premise Vs Hybrid cloud is grappling issue for the companies.
• It involves a huge Change Management effort
• Some of the common challenges faced during decision making are:
a. Lack of accurate Discovery tools, Analytic tool
b. Lack of application context and information
c. Inaccurate On-Premise Cost Estimates
d. Lack of detail w.r.t GRC, licensing
e. Change Management to embrace Cloud
f. Maintaining the clutter of the customizations
g. Psychological Barriers
h. Security
i. Modifying the Interaction patterns
Next few slides will articulate the use of Analytic Hierarchical Process (AHP) to enable the decision making.
AHP is a structured technique for organizing and analyzing complex decisions, based on mathematics and psychology. It was
developed by Thomas L. Saaty in the 1970s and has been extensively studied and refined since then.
3. Identify the Options and Criteria
• The process begins with identifying available options and hierarchy of the
criterion to making decision
• Criteria can be quantitative and qualitative and can form a multiple levels of
hierarchy
Identified Options
Cloud Solution
On-Premise Solution
4. Identified Hierarchy of Criteria
Business Value (Level 1) Risk Value (Level 1) Technology fitment (Level 1)
Time to Market (Level 2) Lack of Maturity (Level 2) Integration Ease (Level 2)
Migration Cost (Level 2) Immature offering (Level 3) No of External Systems (Level 3)
Business Criticality (Level 2) Lack of standards (Level 3) No. of hardware devices for integration (Level 3)
Cost Savings (Level 2) Compliance (HIIPA, PCI) (Level 3) Well defined integration points (Level 3)
Opex Saving (Level 3) Loss of Control (Level 2) Migration Ease (Level 2)
H/W Cost (Level 4) Lack of governance (Level 3) DB Size (Level 3)
S/W Cost (Level 4) Enterprise Policies (Level 3) Application Size (Level 3)
Power Cost (Level 4) SLA (Level 2) Non-Proprietory tool (Level 3)
Administration Cost (Level 4) Security (Level 2) Functionality Complexity (Level 3)
Labor Cost (Level 4) Data Isolation (Level 3) Technology Stack (Level 2)
Capex Savings (Level 3) Data Protection (Level 3) Database (Level 3)
S/W Procurement Cost (Level 4) Lack of audit (Level 3) Runtime (Level 3)
H/W Procurement Cost (Level 4) Data Encryption (Level 3) OS (Level 3)
Application Design (Level 2)
Service Base design (Level 3)
Users virtualization (Level 3)
Note:
For the simplicity of understanding the process, we will deal with Single Level Criteria Hierarchy and Only 2 options.
5. Ranking the Criteria to arrive at the Weightage
Intensity of importance 1 3 5 7 9 2,4,6,8
Definition Equally important
Somewhat more
important Much more important Very important
Extremely
important
intermediate values.
Compromise needed
Select Application
Business Value Risk Value Technology FitmentCriteria
Cloud On-PremiseOptions
Scale Guidelines before ranking the criteria over one other
6. Ranking the Criteria to arrive at the Weightage
Business Value Risk Value Technology fitment
Business Value 1 5 3
Risk Value 0.2 1 5
Technology fitment 0.333333333 0.2 1
Sum 1.533333333 6.2 9
Understanding the Ranking depicted above
1. All the diagonal columns (R1C1, R2C2, R3C3) will be 1 as Business Value over Business value is Equal Importance and so as Risk Value
Over Risk value and Technology fitment over Technology fitment
2. We weigh Business value More Important over Risk Value so R1C2 is 5 and R2C1 is inverse of R1C2 (1/5=0.2)
C1 C2 C3
R1
R2
R3
7. Priority Vector for the Criterion
Business Value Risk Value Technology fitment
Business Value 1 5 3
Risk Value 0.2 1 5
Technology fitment 0.333333333 0.2 1
Column Sum 1.533333333 6.2 9
C1 C2 C3
R1
R2
R3
As next steps, to derive the Priority Vector for all the criterion, we will divide all the individual cells to the Column Sum and Average the Row
value as depicted below
Business Value Risk Value Technology fitment
Priority Vector
(Average of Row Values)
Business Value 0.652173913 0.806451613 0.333333333 0.59731962
Risk Value 0.130434783 0.161290323 0.555555556 0.282426887
Technology fitment 0.217391304 0.032258065 0.111111111 0.120253493
This concludes that the Weightage of the criteria is as follows:
Business Value : 59.73%
Risk Value : 28.24% (+)
Technology Fitment : 12.03%
100.00%
8. Rank Options against the Criteria
• Now we will rank the options against each criteria along the same
lines as illustrated in Slide#6 and 7
Business Value Cloud On-Prem
Priority Vector
(Average of Row values)
Cloud 1 3 0.75
On-Prem 0.333333333 1 0.25
Column Sum 1.333333333 4
Technology fitment Cloud On-Prem
Priority Vector
(Average of Row values)
Cloud 1 6 0.86
On-Prem 0.166666667 1 0.14
Column Sum 1.166666667 7
Risk value Cloud On-Prem
Priority Vector
(Average of Row values)
Cloud 1 0.2 0.17
On-Prem 5 1 0.83
Column Sum 6 1.2
9. Matrix Multiplication for Options and
Criterion weightage
Technology Fitment
(Priority Vector)
Risk Value
(Priority Vector)
Business Value
(Priority Vector)
Cloud 0.86 0.17 0.75
On-Prem 0.14 0.83 0.25
Business Value (Priority Vector) 0.59732
Risk Value (Priority Vector) 0.282427
Technology Fitment (Priority Vector) 0.120253
0.86 0.17 0.75
0.14 0.83 0.25
0.59732
0.282427
0.120253X = 0.651898
0.348102
Conclusion: By doing matrix multiplications, we deduce that Cloud is 65% more preferred over 35%
On-Premise Solution