- Starting date: as soon as possible
- End date: October 1st 2013
In ConPaaS, applications are organized as a collection of services. ConPaaS currently provide services for web hosting (PHP and Java), SQL and NoSQL databases (MySQL and Scalaris), data storage (XtreemFS) and for large scale data processing (Task Farming and MapReduce). Using these services a bioinformatics application could, for example, be composed of a MapReduce service backend to process genomic data, as well as a Web hosting and a SQL database service to provide a Web-based graphical interface to the users. Each service can be scaled on demand to adjust the quantity of computing resources to the capacity needs of the application. A first full version of ConPaaS is already running, and we are now expanding the system with additional functionality.
The postdoctoral researcher will be in charge of the design and implementation of automatic resource provisioning capabilities within ConPaaS. Instead of manually defining how many computing resources should be used to host an application, the user should be able to specify the performance she expects - and trust the platform to continuously adjust the quantity of resources to maintain performance within the requirements while using as few resources as possible.
The main scientific challenge is that applications are not monolithic: if the bioinformatics application described above is not responding fast enough, it is in general not obvious to decide which of the involved services constitutes the bottleneck and should be reprovisioned. We have prior experience with resource provisioning for Web-based applications but now we want to expand the system to more complex applications using, for example, a MapReduce of TaskFarming computing backend.
Besides the activities strictly related to research, the postdoc will help with the management of the ConPaaS team: helping younger researchers, supervising collaborative development, and actively supporting ConPaaS users inside and outside the Contrail project.