Thoughts on the Future of Cloud Orchestration Tools
Jörg Domaschka
a,
Frank Griesinger
a, Daniel Baur
a, Alessandro Rossini
baOMI, University of Ulm, Germany bSINTEF, Oslo, Norway
Targeted shortcomings
1)
Communication description abilities
What we want to do
…
Load balancer
Application server (blog)
Deploying in the Cloud
Deploying an application in the Cloud
:
1) Describe the application
Application
Deploying in the Cloud
Deploying an application in the Cloud
:
1) Describe the application
Component A
Component B Communication
Deploying in the Cloud
Deploying an application in the Cloud
:
1) Describe the application
Deploying in the Cloud
Deploying an application in the Cloud
:
1) Describe the application
2) Click the
green button
Deploying in the Cloud
Deploying an application in the Cloud
:
1) Describe the application
2) Click the
green button
3) Enjoy your application in the Cloud!
Component instance 1 Component instance 2 Component instance 3 install start stop shutdown initialize
Application in the Cloud
Load balancer Application server (blog) DBMS Component Communication Type level Blog applicationApplication in the Cloud
Load balancer Application server (blog) DBMS Component Communication Type level Deploy Blog applicationApplication in the Cloud
Load balancer Application server (blog) DBMS Component Communication Type level Deploy Load balancer Application server (blog) Application server (blog) DBMS…
Component instance Instance level Channel instance Blog applicationCloud Orchestration Tools
Upcoming publication at UCC 2015 by Daniel Baur et al.
•
cloud deployment
•
life-cycle handling
•
monitoring
•
application description
•
…
infinite ca
rdinality
Generic Communication Patterns
Generic Communication Patterns
Generic Communication Patterns
Generic Communication Patterns
Advanced Application Deployment
Assumption: Best performance if each DBMS interface is addressed by a maximum of 2 blog applications. è 1-M mapping (M=2) Load balancer Application server (blog) DBMS Deploy
Advanced Application Deployment
Assumption: Best performance if each DBMS interface is addressed by a maximum of 2 blog applications. è 1-M mapping (M=2) Load balancer Application server (blog) DBMS Deploy Load balancer Application server (blog) Application server (blog) DBMS DBMS Application server (blog)
Port and Link Boundaries
A
à
B : ([ out
low, out
up],[ in
low, in
up])
Port and Link Boundaries
A
à
B : ([ out
low, out
up],[ in
low, in
up])
A
B
Out:
Low .. Up
In:
Low .. Up
Port and Link Boundaries
A
à
B : ([ out
low, out
up],[ in
low, in
up])
Differentiation
:
-
Component
-
Application
-
Application instance
A
B
Out:
Low .. Up
In:
Low .. Up
Port and Link Boundaries
A
à
B
Component: -
Application: [ 1 , 1 ] , [ 1 , 2 ]
Application instance: -
Load balancer Application server (blog) DBMS
A
B
Where are we?
•
Basic communication patterns
VM 1 VM 4 VM 2 VM 5 VM 3 VM 6
Multi-tenant Deployments
Load balancer Application server (blog) DBMS Deploy Load balancer Application server (blog) DBMS Load balancer Application server (blog) DBMSVM 1 VM 2 VM 4 VM 3