Daplacer

<picture> <source media="(prefers-color-scheme: dark)" srcset="img/daplacer-logo-night.png"> <img src="img/daplacer-logo.png" width="300"/> </picture>

Data-aware service placement and routing in the Cloud-IoT continuum.

DA-Placer methodology is described and assessed in:

Jacopo Massa, Stefano Forti, Antonio Brogi<br> Data-Aware Service Placement in the Cloud-IoT Continuum, <br> Communications in Computer and Information Science, vol 1603 @ SummerSOC, 2022.

If you wish to reuse source code in this repo, or mention our work, please consider citing the above mentioned article:

@inproceedings{massa2022daplacer,
  title        = {{Data-Aware Service Placement in the Cloud-IoT Continuum}},
  author       = {Massa, Jacopo and Forti, Stefano and Brogi, Antonio},
  year         = 2022,
  booktitle    = {Service-Oriented Computing},
  publisher    = {Springer International Publishing},
  editor       = {Barzen, Johanna and Leymann, Frank and Dustdar, Schahram},
  pages        = {139--158}
}

<hr>

DA-Placer is a tool for designing and proposing an eligible placement and a suitable routing strategy for a given service-oriented application within its data and requirements, over a given Cloud-IoT infrastructure.

It exploits the backtracking of the Prolog reasoner to entirely explore the search space of possible solutions, and a continuous reasoning approach to speed-up the decision making process at runtime, after a first eligible solution has been found.

As it is written in Prolog, it is very declarative and concise (≃ 170 lines of code), which allows to have a high level of readability and maintainability.

Files  <picture><source media="(prefers-color-scheme: dark)" srcset="https://cdn-icons-png.flaticon.com/512/2822/2822755.png"><img width="20" height="20" alt="files" src="https://cdn-icons-png.flaticon.com/512/2822/2822584.png"/>

</picture> - `model.pl` contains the way to describe the application and the infrastructure on which you want to attempt the placement and routing process.

• data/app.pl and data/infra.pl files contain an instance of the model.

• daplacer.pl contains the main logic of the placer.

How To  <picture><source media="(prefers-color-scheme: dark)" srcset="https://cdn-icons-png.flaticon.com/512/2666/2666505.png"><img width="20" height="20" alt="files" src="https://cdn-icons-png.flaticon.com/512/2666/2666469.png">

</picture>

1. Download or clone this repo. Make sure you have the following prerequisites:

   • swipl
   • python >= 3.6

2. In the project folder, run the following command on terminal, to access the SWI-Prolog console:

   swipl daplacer.pl
   

3. Inside the console, run the following query:

   :- daplacer(museuMonitor, Placement, Routes).
   

   The output will be a first placement and routing for the application described in app.pl (named museuMonitor), onto the infrastructure described in infra.pl.

4. To try the continuous reasoning, open infra.pl file and change some links or nodes involved in the first placement, or change app.pl description, adding new services, data types or end-to-end interactions.

5. Repeat step 2, and you'll obtain as output a new placement for new and suffering services.

   N.B. Do not terminate the already started swipl process, or you will lose the knoweldge base that has been built so far!

Output Example  <picture><source media="(prefers-color-scheme: dark)" srcset="https://cdn-icons-png.flaticon.com/512/3488/3488340.png"><img width="20" height="20" alt="files" src="https://cdn-icons-png.flaticon.com/512/3488/3488804.png"></picture>

% on(Service, Node)
Placement = [on(dataStorage, isp), 
             on(controller, lifeSciences), 
             on(interface, mannLab)],

% ((source, target), AllocatedBandwidth, Route)
Routes = [((rVst, dataStorage), 24.0, [sciencesLectureHall, briggsHall, mannLab, parkingServices, westEntry, firePolice, isp]),  
          ((rArt, dataStorage), 15.0, [kleiberHall, hoaglandAnnex, firePolice, isp]),  
          ((dataStorage, controller), 18.0, [isp, firePolice, westEntry, mannLab, lifeSciences]),  
          ((controller, rDor), 22.5, [lifeSciences, parkingServices]),  
          ((controller, rGls), 27.0, [lifeSciences, parkingServices, mannLab, briggsHall, studentCenter]),  
          ((rCam, interface), 40, [hoaglandAnnex, mannLab]),  
          ((interface, controller), 40, [mannLab, westEntry, parkingServices, lifeSciences]),  
          ((interface, rVid), 20, [mannLab, parkingServices, westEntry])]

The previous prolog output is depicted in the following image:

<kbd><img src="img/result.png" alt="output"/></kbd>

Contributors  <picture><source media="(prefers-color-scheme: dark)" srcset="https://cdn-icons-png.flaticon.com/512/3369/3369157.png"><img width="20" height="20" alt="files" src="https://cdn-icons-png.flaticon.com/512/3369/3369137.png"></picture>

• Jacopo Massa (corresponding author)
• Stefano Forti
• Antonio Brogi