CS2125 Paper Review Form - Winter 2018 Reviewer: Azadeh Assadi Paper Title: CloudMF: Applying MDE to Tame the Complexity of Managing Multi-Could Applications Author(s): Nicolas Ferry, Hui Song, Alessandro Rossini, Franck Chauvel, Arnor Solberg 1) Is the paper technically correct? [X] Yes [ ] Mostly (minor flaws, but mostly solid) [ ] No 2) Originality [ ] Very good (very novel, trailblazing work) [X] Good [ ] Marginal (very incremental) [ ] Poor (little or nothing that is new) 3) Technical Depth [ ] Very good (comparable to best conference papers) [X] Good (comparable to typical conference papers) [ ] Marginal depth [ ] Little or no depth 4) Impact/Significance [ ] Very significant [X] Significant [ ] Marginal significance. [ ] Little or no significance. 5) Presentation [ ] Very well written [X] Generally well written [ ] Readable [ ] Needs considerable work [ ] Unacceptably bad 6) Overall Rating [ ] Strong accept (award quality) [X] Accept (high quality - would argue for acceptance) [ ] Weak Accept (borderline, but lean towards acceptance) [ ] Weak Reject (not sure why this paper was published) 7) Summary of the paper's main contribution and rationale for your recommendation. (1-2 paragraphs) The authors in this paper explain their modeling system aimed at facilitating the deployment of multi-cloud applications. The authors first begin by explaining the issues related with multi-cloud systems, particularly those that run on different infrastructure and platform as a service (IaaS and PaaS). They then explain the various features and components of their modeling system. Broadly, the authors propose a system that creates a new layer of processing that communicates between different IaaS and PaaS clouds to facilitate the smooth running of the SaaS (software as a service) applications, irrespective of the specific host. Essentially, CloudMF has three extensions that consist of support for management of multi-cloud applications relaying simultaneously on both IaaS and PaaS, defining graphical and textual syntaxes for their modeling language, and supporting remote access to the models@runtime by third parties. For the CPIM (provider independent) to CPSM (provider specific) transformation (i.e. to allow an application to function on any cloud provider), this application specifies the cloud provider on which the application will be provisioned and deployed, then, the models@runtime request the cloud providers for a list of available VMS that meet the defined constraints. The cloud provider responds the models@runtime with the list of VMs and the metadata associated with each VM (similarly with the PaaS) and use this data to refine the CPIM into a CPSM that then allows the deployment of the CPSM. In terms of defining the syntax, the authors dedicate a section reviewing the relevant syntax with some limited examples to illustrate their function. Finally, with respect to Models@runtime, authors define them as patterns that allow dynamic adaptation of a system through an abstract representation of the underlying running system. The dynamic nature of these systems, implies that a change in the running system is automatically reflected in the model. Specifically, this process occurs by matching of the existing CPSM with the target CPSM, matching the similarities and recreating those processes while identifying the differences that would either be removed or added to the target CPSM. Given that this process occurs at run-time, changes occurring during reasoning are not handled by CloudMF. However, CloudMF does allow for synchronisation for remote third parties to adapt the system. This is done through propagation of changes in both directions via notifications and commands. This process was however, not clearly explained in this paper. 8) List 1-3 strengths of the paper. (1-2 sentences each, identified as S1, S2, S3.) S1 – Well structured and generally well written except for a few minor grammatical errors S2 – Good use of diagrams to explain the concepts 9) List 1-3 weaknesses of the paper (1-2 sentences each, identified as W1, W2, W3.) W1 – It would have been more beneficial if there were experimental data included to demonstrate the advantages of their design over existing models. The addition of an additional layer of processing can take some time which should be quantified and compared to existing methods of managing multi-cloud applications. W2 – It would have been more beneficial if there were some background information on the workings of cloud computing before diving into the explanation of their proposed design.