A Cloud-based Architecture for the Cherenkov Telescope Array Observation Simulations: Optimization, Design, and Results

Abstract

Simulating and analysing detailed observations of astrophysical sources for very high energy (VHE) experiments, like the Cherenkov Telescope Array (CTA), can be a demanding task especially in terms of CPU consumption and required storage. In this context, we propose an innovative cloud computing architecture based on Amazon Web Services (AWS) aiming to decrease the amount of time required to simulate and analyse a given field by distributing the workload and exploiting the large computational power offered by AWS. We detail how the various services offered by the Amazon online platform are jointly used in our architecture and we report a comparison of the execution times required for simulating observations of a test source with the CTA, by a single machine and the cloud-based approach. We find that, by using AWS, we can run our simulations more than 2 orders of magnitude faster than by using a general purpose workstation for the same cost. We suggest to consider this method when observations need to be simulated, analysed, and concluded within short timescales.

The Application of Electromagnetic Sensors for Determination of Cherenkov Cone Inside and in the Vicinity of the Detector Volume in Any Environment Known

“…Thus, for three iterations and for p = 25 m, three equidistant cubes will result: the first cube will have the m edge; the second cube m and the third cube m. With the three equidistant cubes, we can obtain 32 (it is a suggestive number, it does not represent the number determined by software) the reception electromagnetic sensors compared to the previous situation, where for a cube with the 500 m edge, we obtained 32,000 the reception electromagnetic sensors (much larger number). It can be deduced that the proposed solution, optimizes the Cherenkov detector and contributes to the reduction of realization costs (in the case of our studies—saline environment) of the Cherenkov detector, in any known environment [ 45 ]. To optimize the Cherenkov detector (a minimum number of electromagnetic reception sensors) for any known environment in which a Cherenkov detector can be realized, the following steps will be performed: the use of a simple device to measure the data collected from to the detection’s elements; achievement of the preliminary measurements for the database; creation of a database with the map of the spatial distribution of the dielectric parameters of the known environments for the Cherenkov detector; placement of the electromagnetic reception sensors in the optimal points determined by the software dedicated for this; minimization of the reception electromagnetic sensors number.…”

“…the first cube will have the m edge; the second cube m and the third cube m. With the three equidistant cubes, we can obtain 32 (it is a suggestive number, it does not represent the number determined by software) the reception electromagnetic sensors compared to the previous situation, where for a cube with the 500 m edge, we obtained 32,000 the reception electromagnetic sensors (much larger number). It can be deduced that the proposed solution, optimizes the Cherenkov detector and contributes to the reduction of realization costs (in the case of our studies—saline environment) of the Cherenkov detector, in any known environment [ 45 ]. To optimize the Cherenkov detector (a minimum number of electromagnetic reception sensors) for any known environment in which a Cherenkov detector can be realized, the following steps will be performed:…”

Many-probes multi-object spatially-resolved analyses of galaxy clusters in the big data era

“…To do so, we are constructing an adequate infrastructure which is based on state-of-the-art providers for data storage (Amazon Web Services, e.g. see Landoni et al [15]) and data processing (Apache Spark [16]). Frameworks like Apache Spark allow us to define very specific functionalities on datasets of large dimensions, and to define computation procedures that can be easily scaled on parallel and distributed architectures.…”

The SOXS scheduler for remote operation at LaSilla:Concept and design

“…The purpose of the replication is especially related to the page, seen by the operator at NTT during night time, which permits to trigger the update (or re-population) of the VES in case of observation condition changes. We are exploring two commercial cloud platform 8,9 vendors (Amazon Web Services and Digital Ocean) by comparing overall costs and reliability.…”