CloudScale: Resolving scalability

Gunnar Brataas
Topics recommended for the 2016-2017 Work Programme: 
  1. Improve the scalability engineering method so that it becomes less costly and easier to use and can be used on a broader scale.
  2. Improve the tools required for scalability engineering in the areas of spotting scalability problems in existing code, but also in analysing models as well as in extracting models from existing code and in capturing resource demands parameters from existing systems.
  3. More case studies demonstrating the benefit, but also cost of scalability engineering. This will increase the knowledge of scalability engineering, so that scalability engineering can be used more widely.
Projects major results: 

CloudScale will make a tool suite enabling:

  1. Improving the scalability and cost-efficiency of existing systems by spotting bottlenecks and anti-patterns in the code, both by analysing the running system (dynamic spotting) as well as by analysing the source code (static spotting).
  2. Analysing the scalability and elasticity at design time.
  3. Extracting models from source code to enable analysis.

We are now busy finishing almost complete tools in all these areas and are also working on validating these tools. In the next and final year we will focus on improving these tools using the input from the validation. Dissemination and exploitation will also get more attention.

Potential exploitation strategy: 

Effective exploitation of the results of the CloudScale project depends on the impact the project can have in the domain of scalability in the Cloud. We have identified four main factors in order to have a sustainable practical impact:

  1. A successful internal exploitation of the results by the consortium partners in order to enhance their own products, provide better services, and advance their research contributions, exemplifies the benefits of applying the CloudScale results.
  2. A sustainable exploitation is only possible if the main scientific results and tools developed in the project are open source, easily accessible and available for commercial use.
  3. Presentation of convincing demonstration and value propositions will demonstrate the applicability of CloudScale results to real business scenarios.
  4. Raising awareness for CloudScale together with contributions to international standardisation in this domain play a crucial role in sustainable exploitation of the results of the project.
An update since the last Concertation meeting (March 2014): 

Since the last meeting CloudScale has been active in the following project collaboration and standards related groups:

  • LIMBO by Universitat Wurzburg is used as part of our usage evolution tool.
  • Open Cloud Directory where we have added CloudScale’s showcase CloudStore:
  • SPEC RG Cloud Group: where we presented a live demonstration of CloudStore at their annual meeting on 26-27 March in Dublin. The RG showed interest and considered using CloudStore as well.
  • Palladio Community where we promote our open source strategy using Git as well as CloudStore.
  • FZI: (“Forschungszentrum Informatik”; German technology-transfer institute). We convinced the FZI to build on top of CloudScale’s Analyser in the context of a European project related to energy-efficiency (“CACTOS – Context-Aware Cloud Topology Optimisation and Simulation”).