Section: New Results

Cloud and Edge processing

Benchmarking Edge processing frameworks

Participants : Pedro de Souza Bento Da Silva, Alexandru Costan, Gabriel Antoniu.

With the spectacular growth of the Internet of Things, edge processing emerged as a relevant means to offload data processing and analytics from centralized Clouds to the devices that serve as data sources (often provided with some processing capabilities). While a large plethora of frameworks for edge processing were recently proposed, the distributed systems community has no clear means today to discriminate between them. Some preliminary surveys exist, focusing on a feature-based comparison.

We claim that a step further is needed, to enable a performance-based comparison. To this purpose, the definition of a benchmark is a necessity. We make a step towards the definition of a methodology for benchmarking Edge processing frameworks [20].

Analytical models for performance evaluation of stream processing

Participants : José Aguilar Canepa, Pedro de Souza Bento Da Silva, Alexandru Costan, Gabriel Antoniu.

One of the challenges of enabling the Edge computing paradigm is to identify the situations and scenarios in which Edge processing is suitable to be applied. To this end, applications can be modeled as a graph consisting of tasks as nodes and data dependencies between them as edges. The problem comes down to deploying the application graph onto the network graph, that is, operators need to be put on machines, and finding the optimal cut in the graph between the Edge and Cloud resources (i.e., nodes in the network graph).

We have designed an algorithm that finds the optimal execution plan, with a rich cost model that lets users to optimize whichever goal they might be interested in, such as monetary costs, energetic consumption or network traffic, to name a few.

In order to validate the cost model and the effectiveness of the algorithm, a series of experiments were designed using two real-life stream processing applications: a closed-circuit television surveillance system, and an earthquake early warning system.

Two network infrastructures were designed to run the applications. The first one is a state-of-art infrastructure where all processing is done on the Cloud to serve as benchmark. The second one is an infrastructure produced by the algorithm. Both scenarios were executed on the Grid’5000. Several experiments are currently underway. The trade-offs of executing Cloud/Edge workloads with this model were published in [19].

Modeling smart cities applications

Participants : Edgar Romo Montiel, Pedro de Souza Bento Da Silva, Alexandru Costan, Gabriel Antoniu.

Smart City applications have particular characteristics in terms of data processing and storage, which need to be taken into account by the underlying serving layers. The objective of this new activity is to devise clear models of the data handled by such applications. The data characteristics and the processing requirements does not have to match one-to-one. In some cases, some particular types of data might need one or more types of processing, depending on the use case. For example, small and fast data coming from sensors do not always have to be processed in real-time, but they could also be processed in a batch manner at a later stage.

This activity is the namely the topic of the SmartFastData associated team with the Instituto Politécnico Nacional of Mexico.

In a first phase, we focused on modeling the stream rates of data from sets of sensors in Smart Cities, specifically, from vehicles inside a closed coverage area. Those vehicles are connected in a V2I VANET, and they interact to applications in the Cloud such as traffic reports, navigation apps, multimedia downloading etc. This led to the design of a mathematical model to predict the time that a mobile sensor resides within a geographical designated area.

The proposed model uses Coxian distributions to estimate the time a vehicle requests Cloud services, so that the core challenge is to adjust their parameters. It was achieved by validating the model against real-life data traces from the City of Luxembourg, through extensive experiments on the Grid’5000.

Next, these models were used to estimate the resources needed in the Cloud (or at the Edge) in order to process the whole stream of data. We designed an auto-Scaling module able to adapt the resources with respect to the load. Using the Grid’5000, we evaluated the various possibility to place the prediction module: (i) at the Edge, close to data with less accuracy but faster results; or (ii) in the Cloud, with higher accuracy due to the global data, but higher latency as well.