Architectural Tactics Support in Cloud Computing Providers: The Jelastic Case
1. Architectural Tactics Support in
Cloud Computing Providers:
The Jelastic Case
Jaime Chavarriaga, jchavarr@vub.ac.be
Carlos Noguera, cnogera@vub.ac.be
Rubby Casallas, rcasalla@uniandes.edu.o
Viviane Jonckers, vejoncke@vub.ac.be
2. PaaS providers
PaaS = Platform as a Service
No need to install
infrastructure
Options for:
• Scalability
• Elasticity
• Fault Tolerance
3. From the architect’s point of view…
Before/During deployment, the architect
must configure the PaaS
to achieve the intended QAs
4. e.g. Cloud Platform
Which Environments
(servers) to use?
How many servers ?
Which options
configure in the
environments?
5. Our Contribution
• To provide better abstractions to support
architects configuring PaaS applications based
on:
– Quality Attributes (QAs)
• Architectural tactics
–PaaS Configuration Options
6. Our Approach
• Our solution is based on:
– Feature models to represent variability in:
• Architectural Tactics related to QAs,
• Jelastic configuration options
– Feature Solution Graphs to represent:
• Relationships between these feature models
28. Conclusions
• We have used FM to model variability in architectural
tactics and cloud platform configuration, and FS-Graphs
to model the relationships between them.
• We use these models to help software architects to
configure a PaaS based on QAs and detect conflicts that
may result.
• The same approach can be applied to other cloud
computing provider.
– The corresponding models should be defined using the
knowledge of each specific PaaS provider.
I am
Doing a Phd research in Uniandes Bogota and VUB, Brussels
In the context of variability management
Today I will present an application of our ideas in the context of …
First at all, I will explain the problem of using Paas Providers from the point of view of a software architects
In the context of cloud applications, architects may use…
PaaS providers offer Platform as a Service in cloud computing enviromnents.
A software architects willing to deploy an application may use the execution environments and applications servers offered by a Paas providers such as Google App Engine, Jelastic or providers offering the GRAILS platform.
With these providers, Software architects do not need to install the infrastructure software.
They only has to deploy the application into preinstalled environments.
In addition, they can use additional options for automatic scalability, elasticity and fault tolerance.
From the architects’ point of view, the architect must configure the PaaS environment to achieve the intended Quality attributes for an application.
For instance, if the application must achieve high availability or high performance, the architect must select the proper execution environment, probably using options for multiple servers, clustering or distributed caching systems.
Although the configuration must respond to QAs, the PaaS is configured throught options of the infrastructure
For instance, in Jelastic there are options to configure which brand of application server to use, how many servers to use, and which options include with these servers.
This may result very complex for the architect used to deal with QAs and not with infrastructure options and configuration.
Then, our contribution is focused on helping software architects to configure PaaS platforms based on QAs.
In concrete we are interested on better abstractions to support software software architects configuring PaaS platform based on Quality Attributes.
Basically abstractions that relate Quality Attributes with the architectural tactics that help to achieve these attributes and the Jelastic configuration that implement these architectural tactics.
Now i will present, how we build the solution and some of the results we found.
Feature Models have been used to represent variability in a domain. For instance, to represent which options or configuration elements can be included in an application.
Our approach is based on feature models to represent alternatives about the architectural tactics that can be used to achieve a Quality Attribute and the Diverse options to configure Jelastic.
We are proposing the use of Feature-Solution graphs to relate the architectural tactics to the jelastic configuration options. We use these graphs to represent how an architectural tactic can be implemented using that configuration options.
Extractos
-números, con tamaño
Extractos
-números, con tamaño
Numero de opciones
Note they are features…
From 2,147,483,647 combinations of tactics
only 210,862,079 results in valid configurations in Jelastic.
Future work:
Recommenders for cloud configuration in Jelastic and other platforms
- Tool support for FS-Graphs