The flourishing of Large Language Models (LLMs) calls for increasingly ultra-scale training. In this paper, we share our experience in designing, deploying, and operating our novel Astral datacenter infrastructure, along with operational lessons and evolutionary insights gained from its production use. Astral has three important innovations: (i) a same-rail interconnection network architecture on tier-2, which enables the scaling of LLM training. To physically deploy this high-density infrastructure, we introduce a distributed high-voltage direct current power system and a new air-liquid integrated cooling system. (ii) a full-stack monitoring system featuring cross-host and hierarchical logging correlation, which diagnoses failures at scale and precisely localizes root causes. (iii) an operator-granular forecasting component Seer that efficiently generates operator execution timelines with acceptable accuracy, aiding in fault diagnosis, model tuning, and network architecture upgrading. Astral infrastructure has been gradually deployed over 18 months, supporting LLM training and inference for multiple customers.

As one of the core techniques in 5G, the Internet of Things is more interested than ever. Furthermore, radio frequency identification RFID plays a crucial role in Internet of Things development. Although the low-cost RFID system has wide prospect, it has to face with huge challenges because of potential security risks, privacy problems, and efficiency because of its restrictions on processing, storage, and power in RFID tags. One of the possible solutions in secure authentication of the low-cost RFID system is the lightweight RFID authentication protocol. A lightweight RFID mutual authentication protocol with cache in the reader is proposed in this paper, named LRMAPC. The LRMAPC can greatly reduce the computational and transmission cost. Especially, it can reduce computational costs greatly when a large number of tags want to be authenticated. We prove the correctness of LRMAPC using GNY logic. Compared with some existing works, LRMAPC achieves higher efficiency and stronger security. Furthermore, we developed LRMAPC into ULRMAPC, an ultralightweight RFID mutual authentication protocol with cache in the reader. Compared with SASI and Gossamer protocols, ULRMAPC also achieves higher efficiency and stronger security in storage and computation cost. Copyright © 2015 John Wiley & Sons, Ltd.