The document summarizes a proposal from Turbonomic to help Geico reduce cloud costs on Azure. Turbonomic's solution would analyze Geico's 5,580 ARM VMs currently costing $1.5M per month on Azure. Turbonomic identifies it can generate 3,755 scaling actions to optimize resources, reducing monthly costs by $575K (38%) through rightsizing, deleting unused resources, and optimizing reserved instances. The proposal outlines a plan to implement Turbonomic in phases to initially prove savings, then integrate automation and expand across Geico to fully realize the cost savings and operational efficiencies Turbonomic's continuous optimization can provide.
2. Proof of Value Results
1. Only Turbonomic Solves Geico’s Problem
2. Current Mode of Operations
3. Current Analysis
4. Our Plan
5. Financial Impact and Model
6. Appendix
3. Turbonomic: A Strategic Partner for Geico’s Cloud Journey
Why Application Resource Management?:
• Reduce Cost: Geico has a best in class RI coverage yet there is great potential to further reduce cost by properly
allocating resources to applications by de allocating over provisioned and idle resources.
• Continuously Assure Performance: Demand of applications is always changing. You need a real-time system to
continually recognize these changes in demand and appropriately allocate the underlying configuration.
• Keep up with change: Azure constantly releases new SKUs that are better performing and/or cheaper.
Why Turbonomic? :
• Application driven: Turbonomic is the ONLY platform able to understand applications requirements and all tradeoffs
in real-time, ensuring that all generated actions are safe to execute and will not negatively impact app performance.
• Scalable: Relying on recommendations that require people to review and decide how to react does not scale to
Geico’s needs. Turbonomic’s trustworthy actions allow Geico to build a highly scalable practice.
• Automation: The only way to truly reduce cost at scale is by actioning the decisions automatically. Turbonomic is the
only platform that generates trustworthy decisions that do not rely on human oversight and can be automated.
Why Now? :
• Right now there are a total of 3,700+ scaling actions which will drive the environment to a more optimal state.
• CWOM can generate a savings of $575K per Month on the workloads (5,580) covered in the pilot.
• Current practices do not scale
4. Current Mode of Operations
1. GEICO has deployed over 5,580 ARM VMs in Azure
2. Costs are approximately $1.5M/mo
3. Current RI coverage model is at a best in class 90%+
4. Cost is still a concern organizationally
5. RI Purchasing is manual, time intensive and allocation based vs. consumption based
5. Current Analysis
Reserved Instances were purchased by Geico Team on 4 March
• These RI’s were purchased based on Current sizing of VMs
• This purchase reduced the monthly bill by $200k ($1.7M to $1.5M)
93% of Workloads Covered
with RI’S!
6. Current Analysis
• 93% RI Coverage is Awesome…
• It’s in the Top 5% of all of our customers
• But it locks in your resources and costs per those workloads
• What if we could “Right Size” your RI’s and scale all other compute and storage resources?
• Scaling actions would TRIPLE from 1,140 to 3,755
• 70%+ of Workloads will be resized
• Savings would be $575k/Month
7. How to get from $1.5M down to $900K
1. If GEICO had NO RI’s purchased, the current cost for Azure VMs would be - $1.9M
2. If GEICO had right sized based without memory metrics the monthly cost would be - $1.4M
◦ This savings is ~70K MORE than using RIs
3. If GEICO had enabled memory metrics - $1.1M
4. If GEICO then bought RI’s after right sizing cost would be - $1M
5. Additional savings could be driven through suspension and deleting unattached storage
bringing costs to $900K
38% Decrease
on 5,580 VMs
8. Centralized approach
Best for centralized provider models
Has single team covering all LOB’s
Initiatives run in parallel across LOB’s
Progression dictated by slowest LOB
More of a waterfall approach
Decentralized approach
Suites independent LOB’s model
Multiple teams, each aligned to LOB
Initiatives are run in LOB verticals
Proceeds at asymmetrical pace across LOB’s
More competitive
Better supports a sprint-type methodology
How Do We Get to Value Realization?
9. Quick ROI
Initial Automation
Adoption & Integration
Expanded Automation
IT Transformation
Reclaim cloud
inefficient resources
Manually scale pre-
selected applications
not covered by RIs
Show saving and value
with Personalized LOB
views
Identify Directors to
exercise initial workload
optimization
Execute non-prod actions
in one-time change
windows to achieve
savings
Align Scaling Actions with
RI Exchanges
Employ lessons learned
Expand to additional LOBs
and recurring change
windows
ServiceNow &
AppInsights Integration
Enable automation for
app performance
improvements
Expand activities to
automate production
IaaS to PaaS
Azure DevOps + Azure
ARM Templates +
CI/CD Pipeline
Integration
Expand and align with
advanced use cases
Deliver advanced
Application Resource
Management
Align with modernization
efforts – Containers, etc.
BUILD TRUST
ADVANCE ADOPTION
DELIVER ARM
Near term goals
1H 2020
Long term goals
2H 2020
The Proposed Value Realization Plan
10. $1.7M
Achieved Cloud Savings Executed Actions
244
On-Boarded Apps
14
19,300 VMs
137 Subscriptions
2,300 DBs
$23.3M
Identified Savings
“We have lots monitoring tools but in terms
of giving me the exact recommendations
what do I need to do and what actions to
take, there is nothing like that - expect for
Turbonomic. Humans are not sustainable
for optimization, it requires Turbonomic“
Start with quick ROI projects…
Director,
Global Cloud
Adoption
Top 5
Consulting
Company A
11. $260K
Achieved Cloud Savings Executed Actions
511
Over a single day
“The app owners initially pushed back about not
wanting to scale down their workloads, but after
we did the scaling exercise, we didn't hear a peep
from anyone about performance issues, and we
reduced our cost"
Extend to initial automation…
Large Pharmaceutical A
Director, IS
Tech-OPS
12. Geico Automation Workflows
Continuous Resource
Management
Resize/ Config/
Move/Start/Delete
Actions
Change Management
Create ticket for User
Approvals
CI Creation, Action Execution, CI Update
Action Orchestration
Execute Approved Action
Audit Log
Close ticket and
track all executed
changes
Application Monitoring
Collect metrics and group
business applications
Cloud Providers
API call to Log and
Notify Turbo of
Change
Track and keep all
executed actions
13. “A lot of VMs are overprovisioned and we pay a
premium for things we don't need, Turbonomic
shines the light on these inefficiencies”
$12M
Total Cloud Savings to-date
745
Performance Issues
Resolved/month
13K
Savings Actions/month
Expand adoption and integration…
Top 5
Consulting
Company B
Associate
Director,
Infrastructure &
Cloud Services
14. $466K
Cloud Savings/month
Scaling Cloud Executed Optimizations
468
(81%)
Cloud Savings/month
Unattached Volume
$3K
$339.36 $321.41
$297.88
$328.74 $323.77
$285.05 $276.80 $278.34 $263.54
$239.78
$-
$100.00
$200.00
$300.00
$400.00
Feb Mar Apr May Jun Aug Sep Oct Nov Dec
Average Cost Of Cloud VM ($/Month)
Total
Decreased 30%
“The program we built around
“Allergan’s automation engine”
allows us to be proactive in how
we manage our cloud estate.
There is nothing else like it.
Expanded Automation… IT Associate
Vice President
Large Pharmaceutical B
15. ROI
Inputs:
Savings: 34%, or $559K monthly-10%
growth in instances YoY
Scope: 5,580 instances
Licensing Cost: $1.93M, 3 Year paid
upfront
Recommended Services:
~$198k Yr 1
~$89k Yr 2
~$89k Yr 3
(requires scoping session)
Financial Impact
Break Even in 9 Months
16
These are conservative
estimates based on our
ramped PS plan
$-
$5,000,000.00
$10,000,000.00
$15,000,000.00
$20,000,000.00
$25,000,000.00
1 2 3 4 5 6 7 8 9 101112131415161718192021222324252627282930313233343536
Savings
Months
3 Year ROI
Investment Total Savings
$20M+ Savings
over 3 years
16. Key Points for Turbonomic
Control Cloud Costs
Safely reduce monthly Azure bill by
35-50%
Integrated with Automation Pipeline
API-driven, agentless system
integrates with Azure DevOps,
AppInsights, ServiceNow
Automated
Decision making and integrated
processes to execute trustworthy
actions (right-sizing, deleting,
suspending) reduce the manual
effort required by over 50%
Future Proof
Real-time decision engine that is holistic
and understands the entire
environment and optimizes containers
across any cloud
(AWS, Azure, GCP)
RI-Aware Right-Sizing
Right-sizing optimization decisions
consider existing RI inventory and will
maximize utilization of existing RI’s
Dashboards and Reporting for App
Owners, Execs, and Admins
Customizable policies and views for
each business unit to consider realistic,
executable actions only. Also track
savings identified vs. realized.
Migration to Cloud Planning
Accurately plan on-premises to
cloud migrations efficiently to avoid
overspending from day 1
Applications First
Execute actions that do not
negatively impact app
performance
Size Up for Performance
Build trust with app owners to give apps
the resources they need – when they
need them, including size up for better
performance
Editor's Notes
Self service for app owners to own the optimization
Track actions accepted, savings
RI tracking via linked accounts vs. global account
Real-time vs. monthly reports
CI/CD automation pipeline to accelerate BU approvals
Wasted storage is any disk space devoted to files that are not required for operations of the workloads in your cloud environment. It can indicate opportunities for you to free up disk space, and reduce your overall cloud costs.