China
Africa
Explore chapters and articles related to this topic
Aircraft
Published in Suzanne K. Kearns, Fundamentals of International Aviation, 2018
Aviation companies try to optimize the use of each of their aircraft – they don’t maintain several back-up aircraft that sit idly by in case one aircraft in the fleet requires extra time at the AMO/MRO. Therefore, when unscheduled maintenance is required, the removal of an aircraft from service can cause wide-ranging scheduling problems that can impact an entire organization. Aviation companies try to mitigate these disruptions by employing: A minimum equipment list (MEL) – the MEL identifies the essential equipment that must be functional for the aircraft to remain in service. If a component is inoperative, but is not crucial for the flight and therefore not on the MEL, the aircraft can safely remain in service and the repair will be deferred to a later time. A secondary back-up instrument in the cockpit is an example of a repair that can be deferred because the aircraft can safely fly without it. By contrast, a functional fuel gauge would be on the MEL – an aircraft would have to be removed from service immediately so that it could be repaired.Line replaceable units (LRU) – a piece of equipment that has been designed to be removed (and replaced quickly and easily) in case of failure to minimize delay. The transponder in the cockpit is an example of an LRU – it can be replaced quickly using standard connection points and attachments without having to send the aircraft to the hanger for maintenance.Equipment redundancy – some aircraft are built with redundant back-up units, so that if one fails, the other will take over.27 For example, most aircraft have both a primary and a back-up radio in the cockpit.
Improving storage performance of high-performance computing systems by using the non-volatile buffer cache
Published in International Journal of Computers and Applications, 2021
Hyunkyoung Choi, Yong-Hyeon Shin, Hyokyung Bahn
For now, as the capacity of non-volatile memory is limited, a space-efficient management technique for the non-volatile buffer cache is necessary. In this paper, instead of storing an entire block to the non-volatile buffer cache, we only maintain the modified part of a block, thereby improving the space-efficiency of non-volatile memory. Instead, as the modification is also reflected to the volatile buffer cache, all read/write and flush operations can be performed by accessing the volatile buffer cache. Note that the non-volatile buffer cache is used only when a system crash occurs. When free space is necessary in the volatile buffer cache, we use the LRU (least recently used) algorithm to select an eviction victim as LRU is the most popular replacement algorithm used in buffer cache systems. When a modified block is selected as the victim block in volatile buffer cache, it is first written to storage, and then discarded. As all modifications of a block are also maintained in the non-volatile buffer cache, we also remove the fragments of the victim block from the non-volatile buffer cache. Note that these fragments have already been reflected to storage, and do not cause inconsistency problems any more.
Cost aware cache replacement policy in shared last-level cache for hybrid memory based fog computing
Published in Enterprise Information Systems, 2018
Gangyong Jia, Guangjie Han, Hao Wang, Feng Wang
We use Instructions Per Cycle (IPC) as the parameter to measure the cache performance. Figure 6 shows the IPCs under CACRP, HAP, LRU with different capacity configurations. CACRP is the replacement policy proposed in this paper. HAP is the last level cache partition policy for hybrid memory (Wei et al. 2014). LRU is the baseline cache management policy. In the figure, as the capacity ratio between DRAM and NVM is decreased, the cache performance improves. In case the capacity ratio is big, it means that there is more DRAM capacity being used in the main memory. In such cases, the LRU policy is used. Since this policy works well originally, therefore any subsequent performance improvement will be low. On the other hand, if the capacity ratio is low, it means that there is less DRAM capacity being used in the main memory. In such cases, the initial LRU performance is poor, and any subsequent performance improvement will be high. In other word, both CACRP and HAP work better in the case of a hybrid memory system with less DRAM. Current hybrid memory systems have less DRAM and more NVM to decrease the energy consumption making them ideal for the application of CACRP.
An architecture for synchronising cloud file storage and organisation repositories
Published in International Journal of Parallel, Emergent and Distributed Systems, 2019
Gil Andriani, Eduardo Godoy, Guilherme Koslovski, Rafael Obelheiro, Mauricio Pillon
The local cache is populated on demand. The Cloud4NetOrg architecture provides the user with access to the entire set of files linked to their account on the cloud provider. As soon as a read or write operation is triggered, if the file involved is not in the local cache or is outdated, it is transferred from external storage to the local cache. The content replacement algorithm applied to the local cache is LRU (the least recently used files are discarded).