Use Existing Assets to Build a Powerful In-house Cloud Solution - Magali Pervan, Joffrey Luangsaysana, Grégoire Lamodière - DIMSI
- 10. A customer has contacted DIMSI to
make a global analysis of their
server compute usage.
After a move to the cloud end
2020, they lost control on budget
and are no longer able to a have
clear overview of their expenses
(iops, bandwidth, etc)
The Project
❑ There is no devops process, and
every comiter is free to define
his own methodology
❑ They still have legacy hardware
❑ They now want to get control
back on their usages and
expenses
- 11. General informations
❖ Medium sized business in Paris
(100 people)
❖ Digital Strategy consulting
❖ Existing hardware from 2013-2016
❖ Moved their workload to public
cloud between 2018 and 2020
Their Needs
✓ Keep costs under control
✓ Leverage existing Infrastructure
✓ Keep the same flexibility
- 12. Project Team
Grégoire Lamodière
Global Delivery
Magali Pervan
Scrum Master
Joffrey Luangsaysana
Cloud expert
Sophie P.
Partner, head of IT
Jean-Marc D.
Sysadmin
Navid Abdoul
System admin
- 13. Strategy and phases of our work
Initial
audit
Cloud usage
Hardware
Connectivity
Strategy
definition
Solution
implementation
Tests
FAAS
preparation
Go
Live
- 14. June July
2023 2023
5 days
Audit
10 days
Physical Setup and Cloudstack Implementation
7 days
Kubernetes with FaaS Implementation
5 days
Pipelines
5 days
Tests
5 days
Delivery
Planning
Total effort : 37 man.days
- 15. Step 1 : Audit
What are the customer cloud usage ?
▪ Microsoft Azure : Mainly
serverless, Function As a Service
How good is the on-premise infrastructure?
▪ 1G ISP bandwith
▪ 10G switches
▪ SSD and HDD storage arrays (QNAP 10
Gbe)
▪ IBM 3550 M5 with Intel Xeon E5 v4
Redundancy ?
- 17. Step 2 : Strategy
Hybrid Strategy
• Repatriate On Premise Poc, dev environment and
non critical workload to reduce cloud hosting costs
• Keep critical workloads, such as pre-prod and production that need geo clustering,
high level of redundancy and scaling in Azure
▪ Use CloudStack for its IaaS capabilities to leverage the hardware
▪ Use same tech on cloud and on premise: Kubernetes with Azure functions / AWS Lambda
▪ Rely on the Cloudstack Kubernetes service
- 19. ❑
❑
❑
KVM installation (4 x Ubuntu hosts)
❑ 128 Go Mem
❑ 2 x E5v4 2,1 Ghz CPU (total 32 cores)
Step 3 : Hardware preparation & installation
- 23. CI/CD
▪ Smooth delivery from development to production environment
▪ Pipelines from Cloudstack Kubernetes to Azure Cloud / AWS
- 25. Costs and ROI
Quantity Type vCPU Memory Storage Unit. price Total price
8 IAAS Windows
(A8v2)
8 16 Gb 80 Gb 480 € 3 840 €
8 Function App
premium
(EP2)
2 7 Gb 250 Gb 319 € 2 552 €
TOTAL 80 vCPU 184 Go 2,64 Tb ~ 6 500 €
Azure workloads to repatriate to Cloudstack
Name Type Price
Electricity Opex 0,8 k€
Migration project Capex 30 k€
Maintenance Opex 1,2 k€
Windows licence Capex 4 k€
Additionnal onPremise pricing
Server consumption : estimated : 3 Kva
External monitoring : 50 €
Maintenance : 2 man.days / month
Windows Server licence : 500 € / VM
- 26. Costs and ROI
Month 1 Month 2 Month 3 Month 8 Month 12 Month 36
Full cloud 6500 6500 6500 6500 6500 234000
Rollup cost 6500 13000 19500 52000 78000 234000
Hybrid 36000 2000 2000 2000 2000 2000
Rollup cost 36000 38000 40000 50000 58000 106000
Break even : 8 months
Net. saving (36 months) : 128 k€
0,00 €
50 000,00 €
100 000,00 €
150 000,00 €
200 000,00 €
250 000,00 €
hybrid rollup cost Cloud rollup cost