This document discusses approaches to managing applications across multiple cloud platforms, or multi-cloud management. It advocates taking a pragmatic approach using containers, Kubernetes, and automation. Kubernetes can provide a common layer of abstraction across clouds and simplify the multi-cloud challenge. An example approach is presented using ONAP microservices and Kubernetes to deploy applications across clouds. Automation is key to allow integration between clouds and platforms in a hybrid environment. The summary emphasizes that hybrid cloud is a reality and Kubernetes with standards like TOSCA can help provide interoperability when managing applications on multiple clouds.
9. The Challenge
Operational
Chaos:
● Who’s running what?
Where?
● How much does it cost?
● Who is running ON what?
Automation, cost, and
performance optimization are
bound to a specific platform
11. Cloud Management Platform - Infrastructure first
Pros:
● Single Pane of Glass
● Cost analysis and control
Cons
● Limited application
awareness
● Doesn’t fit well with
devops processes
● Limited cloud services (Least
Common Denominator)
12. PaaS - Application First Approach
Pros:
● Kubernetes and containers
are supported by all major
clouds
● Application / Micro Services
centric
Cons
● Suits mostly for Green Field
● Lack of interoperability with
the rest of the world..
13. Multi-Cloud Orchestration - Automation First
Pros:
● Works across a wide variety
of public / private clouds
● Can support any cloud
service or application
● Highly customizable
Cons
● Requires customization per
application / cloud
14. Too Many Options?
14
Your apps
sucks,
rewrite!
No More
VMware!
Docker is
awesome
dude!
Private Cloud
is Dead!
We’ll Solve all
Your
problems..
15. Take the Pragmatic Approach to Multi-Cloud
● The only constant is change!
● There ain’t a single solution
that fits all
● Keep your options open - Avoid
locking into a single platform
16. Take the Pragmatic Approach to Multi-Cloud
● Let’s be more specific ..
○ Use Containers where possible
○ Kubernetes simplify the multi cloud challenge
(Supported by all major clouds)
○ Use Automation Framework to allow
integration and interoperability between the
various clouds and platform
○ Use a declarative approach to abstract the
demand (application), from the supply
(infrastructure resources)
17. Example: A Pragmatic Approach to Multi-Cloud
ONAP Operation Manager (OOM)
● New linux Foundation Project
● Comprised out of Dozen
Micro Services using
Docker/Kubernetes
● Multi-Stack/ Multi Cloud
18. Example: A Pragmatic Approach to Multi-Cloud
Manage Kubernetes Infrastructure
On Multi Cloud
● Create VM/Baremetal
machine
● Configure the
network/Storage, Security,
Monitoring ..
● Seamless integration (Using
the Kubernetes Provider
plugin)
● Auto-Scale, Heal workflow (or
manual)
Kubernetes Nodes
Kubernetes Master
Network
Setup
Kubernetes Multi Cloud Cluster Blueprint
19. Example: A Pragmatic Approach to Multi Cloud
Manage ONAP Microservices on
Kubernetes
● Templatize the standard
Kubernetes template (in-line
or by reference) ..
● Dependency and discovery
between services.
● Allow execution of workflow
Deployment
Blueprint
Plugin
20. Example: A Pragmatic Approach to Multi Cloud
Support Hybrid Stack..
● Support a combination of
Kubernetes services
alongside standard VM’s ,
BigData applications, Lamda
Services, Legacy Application
under the same
automation/orchestration
template
Deployment
Blueprint
Plugin
21. Example: A Pragmatic Approach to Multi-Cloud
Multi Site / Environment..
● Managing Multiple
Kubernetes clusters
○ Cross site deployment
○ Dev, QA, Production
○ Multi-Cloud
Private
Public
Bare-metal
22. Live Demo..
● Define a new service into a
deployment template
● Deploy the service into an
existing Kubernetes cluster.
23. Summary
● Hybrid cloud is a reality you can’t avoid
● Kubernetes provides a good abstraction
for managing application on multi-cloud
● Standard modeling such as TOSCA
makes the interoperability of Kubernetes
with the outside world possible