While there have been important advances within wireless communication technology, the provision of communication support during disaster relief activities remains an open issue. The literature in disaster research reports several major restrictions to conducting first response activities in urban areas, given the limitations of telephone networks and radio systems to provide digital communication in the field. In search-and-rescue operations, the communication requirements are increased, since the first responders need to rely on real-time and reliable communication to perform their activities and coordinate their efforts with other teams. Therefore, these limitations open the door to improvisation during disaster relief efforts. In this paper, we argue that flying ad-hoc networks can provide the communication support needed in these scenarios, and propose a new solution towards that goal. The proposal involves the use of flying witness units, implemented using drones, that act as communication gateways between first responders working at different locations of the affected area. The proposal is named the Flying Real-Time Network, and its feasibility to provide communication in a disaster scenario is shown by presenting both a real-time schedulability analysis of message delivery, as well as simulations of the communication support in a physical scenario inspired by a real incident. The obtained results were highly positive and consistent, therefore this proposal represents a step forward towards the solution of this open issue.

Disaggregated memory is under investigation as a way to break the rigid boundaries between node memory hierarchies in order to provide memory as a system-wide pooled resource. The resource manager allocates the system’s disaggregated memory to jobs, based on the memory requirements defined by the user at job submission time. It is hard for the user to know the job’s precise peak memory footprint, and prior work has shown that users have an incentive to overestimate their needs. This overestimation leads to a significant overallocation of memory, and the majority of the physical memory in the system is wasted. This paper presents a way to reclaim much of this overallocated memory. We extend the Slurm job scheduler to dynamically reallocate memory, according to the job’s current memory footprint. We enhance an existing Slurm simulator to model this situation and combine publicly available traces to model an HPC system on up to 1490 nodes. Our results show that the dynamic memory provisioning approach increases the throughput per dollar by up to 38%, compared to a system with static allocation of disaggregated memory.