How is generative AI impacting your infrastructure?

生成式人工智能如何影響數(shù)據(jù)中心運(yùn)營(yíng)商的基礎(chǔ)設(shè)施？

In the rapidly evolving realm of AI technology, new developments surface nearly every day. Clearly, AI possesses a significant capacity to transform our lives, a technology that spans chatbots, facial recognition, self-driving vehicles, and early disease detection.

在快速發(fā)展的人工智能技術(shù)領(lǐng)域，幾乎每天都有新的進(jìn)展。顯然，人工智能擁有改變我們生活的巨大潛力，這項(xiàng)技術(shù)涵蓋了聊天機(jī)器人、面部識(shí)別、自動(dòng)駕駛汽車和早期疾病檢測(cè)。

The global AI market was valued at $142.3 billion in 2023, with finance, healthcare, and the high-tech/telco markets taking the lead in AI adoption.

2023 年，全球人工智能市值達(dá) 1,423 億美元，其中金融、醫(yī)療保健和高科技/電信市場(chǎng)將率先采用人工智能。

AI is already being used to monitor data center assets, proactively detect faults and improve energy efficiency by driving better power usage effectiveness (PUE). And its not just being used by Hyperscalers, but also by many large enterprise companies.

人工智能已被用于監(jiān)控?cái)?shù)據(jù)中心資產(chǎn)，主動(dòng)檢測(cè)故障，并通過(guò)提高電源使用效率（PUE）來(lái)改善能源效率。而且，這不僅僅是超大規(guī)模數(shù)據(jù)中心才會(huì)使用，它還被許多大型企業(yè)公司所采用。

InfiniBand versus Ethernet

InfiniBand 與以太網(wǎng)對(duì)比

Ethernet remains the prevailing global standard in most data centers. But an increasing number of today’s AI networks now use InfiniBand technology, although InfiniBand holds a mere fraction of the market share at present, primarily for HPC networks.

以太網(wǎng)仍然是大多數(shù)數(shù)據(jù)中心的全球標(biāo)準(zhǔn)。但是，現(xiàn)在越來(lái)越多的人工智能網(wǎng)絡(luò)使用 InfiniBand 技術(shù)，盡管 InfiniBand 目前僅占市場(chǎng)份額的一小部分，主要用于 HPC 網(wǎng)絡(luò)。

Competition is emerging between InfiniBand market leaders and prominent Ethernet switch and chip manufacturers, whose next-generation chips have been designed to construct AI clusters using Ethernet instead of InfiniBand. Regardless of the protocol chosen, both InfiniBand and Ethernet share requirements for high bandwidth and low latency, necessitating top-tier optical cabling solutions for optimal performance to support large language model (LLM) training and inferencing.

InfiniBand市場(chǎng)領(lǐng)先者與知名的以太網(wǎng)交換機(jī)和芯片制造商之間的競(jìng)爭(zhēng)日益激烈，這些制造商的下一代芯片旨在使用以太網(wǎng)而非InfiniBand構(gòu)建的AI集群。無(wú)論采用哪種協(xié)議，InfiniBand和以太網(wǎng)都需具備高帶寬和低延遲，因此需要頂級(jí)的光纜解決方案來(lái)實(shí)現(xiàn)最佳性能，用以支持大型語(yǔ)言模型（LLM）的訓(xùn)練和推理。

Exponential demands for power and bandwidth

對(duì)于電力以及帶寬的指數(shù)級(jí)需求

Two of the key challenges that data centers are facing relate to extreme power needs and associated cooling requirements for the equipment, and the exorbitant bandwidth needs of the GPUs.

數(shù)據(jù)中心正面臨著兩個(gè)關(guān)鍵性挑戰(zhàn)，一個(gè)挑戰(zhàn)是設(shè)備極高的電力需求以及相關(guān)設(shè)備冷卻的需求，另一個(gè)挑戰(zhàn)則是GPU極高帶寬的需求。

Supercomputers with GPUs running AI applications demand vast power and multiple high-bandwidth connections. These GPUs demand from 6.5kW to over 11kW per 6U unit. When contrasted with packed data center cabinets, averaging 7-8kW and maxing at 15-20kW per cabinet, the extent of AI’s power appetite becomes clear. Many of the leading Server OEMs are also offering servers with these GPUs.

配備GPU的超級(jí)計(jì)算機(jī)運(yùn)行人工智能應(yīng)用需要大量的電力以及多個(gè)高帶寬連接。這些GPU每6U個(gè)單位的功耗從6.5千瓦到超過(guò)11千瓦不等。與平均每個(gè)機(jī)柜7-8千瓦，每個(gè)機(jī)柜最大15-20千瓦的密集型數(shù)據(jù)中心機(jī)柜相比，人工智能對(duì)電力的需求顯而易見。許多頭部原始設(shè)備制造商們（OEMs）的服務(wù)器也提供配備這些GPU。

These GPUs typically need connections with bandwidth of up to 8x100Gb/s (EDR), 200Gb/s (HDR) or 400Gb/s (NDR). Every node commonly has eight connections, equating up to 8x400G or 3.2 terabit per node.

這些GPU通常需要帶寬高達(dá)8x100Gb/s（EDR）、200Gb/s（HDR）或400Gb/s（NDR）的連接。每個(gè)節(jié)點(diǎn)通常有八個(gè)連接，相當(dāng)于每個(gè)節(jié)點(diǎn)的總帶寬為8x400G或3.2兆比特。

How will IT infrastructure cope with these requirements?

IT基礎(chǔ)設(shè)施將如何應(yīng)對(duì)這些需求？

Data center power and cooling demands are pushing network managers to reconsider their infrastructure. This often involves altering network blueprints and spacing out GPU cabinets further, potentially adopting end-of-row (EoR) configurations to better handle escalating temperatures.

數(shù)據(jù)中心的電力和冷卻的需求正迫使網(wǎng)絡(luò)管理人員重新考慮他們的基礎(chǔ)設(shè)施。這通常涉及改變網(wǎng)絡(luò)藍(lán)圖以及進(jìn)一步拉大 GPU 機(jī)柜的間距，甚至可能采用行端（EoR）配置，以更好地應(yīng)對(duì)不斷升高的溫度。

However, this means an increased physical gap between switches and GPUs. To accommodate this, data center operators might need to incorporate more fiber cabling used for switch-to-switch connections. Given these extended spans, direct attach cables (DACs) are unlikely to be suitable as they are confined to five meters at most for such speeds.

但是，這也意味著交換機(jī)和GPU之間的物理間隙拉大。為了適應(yīng)這種情況，數(shù)據(jù)中心運(yùn)營(yíng)商可能需要增加光纖布線用于交換機(jī)之間的連接。 鑒于跨度如此之大，直接連接電纜 (DAC) 不太可能適用，因?yàn)樵谶@種速度下，直接連接電纜的長(zhǎng)度最多只能達(dá)到 5 米。

Active optical cables (AOCs) are also a feasible choice thanks to their capacity to cover greater distances compared to DACs. AOCs offer the added advantages of significantly reduced power consumption in comparison with transceivers, as well as enhanced latency.

活動(dòng)光纜（AOCs）也是一種選擇，因?yàn)樗鼈兡軌蚋采w比DACs更遠(yuǎn)的距離。與光收發(fā)器（DACs）相比，活動(dòng)光纜還具備功耗顯著降低，延時(shí)更短等優(yōu)點(diǎn)。

Transitioning data center backbone interconnections between switches will necessitate parallel optic technology to sustain increasing bandwidth demands. Several existing choices for parallel fiber optic technology employ eight fibers in conjunction with multi-fiber push-on connectivity (MPO/MTP fiber connectors). These MPO Base-8 solutions permit the adoption of either singlemode or multimode fiber and facilitate smooth migration to higher speeds. For enterprise data centers, contemplating a Base-8 MPO OM4 cabling solution is advisable when upgrading to 100Gb/s and 400Gb/s. Conversely, cloud data centers should sext a Base-8 MPO singlemode cabling solution while transitioning to 400Gb/s and 800Gb/s speeds.

為了滿足不斷增長(zhǎng)的帶寬需求，數(shù)據(jù)中心骨干間的交換機(jī)之間連接必須采用并行光學(xué)技術(shù)?，F(xiàn)有的幾種并行光纖技術(shù)采用八根光纖與多光纖推入式連接（MPO/MTP 光纖連接器）。這些 MPO Base-8 解決方案允許采用單?；蚨嗄９饫w，便于向更高速度平穩(wěn)遷移。對(duì)于企業(yè)數(shù)據(jù)中心來(lái)說(shuō)，在升級(jí)到 100Gb/s 和 400Gb/s 時(shí)，最好考慮使用 Base-8 MPO OM4 布線解決方案。

Innovative new fiber enclosure systems on the market can flexibly support different fiber modules, including Base-8 and Base-12 with shuttered LC, MTP pass-thru modules, and splicing modules. They allow for easy access and improved cable management.

市場(chǎng)上創(chuàng)新的新型光纖外殼系統(tǒng)可以靈活地支持不同的光纖模塊，包括帶閉合 LC 的 Base-8 和 Base-12、MTP 直通模塊和熔接模塊。它們可以方便地接入并改進(jìn)電纜管理。

In the realm of AI applications, where latency holds immense significance, Siemon suggests opting for “AI-Ready” solutions employing ultra-low loss (ULL) performance alongside MTP/APC connectors. The incorporation of ultra-low-loss fiber connectivity becomes pivotal for emerging short-reach singlemode applications (backing 100, 200, and 400 Gb/s speeds over distances exceeding 100 meters). This ULL connectivity effectively meets the more stringent insertion loss prerequisites set by AI applications, thereby enhancing the entirety of network performance.

在人工智能應(yīng)用領(lǐng)域，時(shí)延具有極其重要的意義，Siemon 建議選擇 "人工智能就緒"（AI-Ready）解決方案，該方案采用超低損耗（ULL）性能和 MTP/APC 連接器。對(duì)于新興的短距離單模應(yīng)用（在超過(guò) 100 米的距離上支持 100、200 和 400 Gb/s 的速度）而言，采用超低損耗光纖連接至關(guān)重要。這種超低損耗連接可有效滿足人工智能應(yīng)用所設(shè)定的更為嚴(yán)格的插入損耗要求，從而提高整個(gè)網(wǎng)絡(luò)的性能。

Additionally, Expert advises the adoption of APC (angled physical connect) fiber connectors, including the MTP/APC variant, for specific multimode cabling applications, alongside the traditional singlemode approach. The angle-polished end-face configuration of APC connectors (in contrast to UPC connectors) reduces reflectance, thus elevating fiber performance.

此外，除去傳統(tǒng)的單模方法外，專家 還建議在特定的多模布線應(yīng)用中采用 APC（帶角度的物理連接）光纖連接器，包括 MTP/APC 變體。APC 連接器（與 UPC 連接器相反）的端面角拋光結(jié)構(gòu)可減少反射，從而提高光纖性能。

AI stands as a disruptive technology, yet it harbors the capacity to transform not just our professional lives but the very fabric of our existence—and data center operators need to prepare for it. Adopting measures to facilitate a seamless shift to elevated data speeds, and enhancing the energy efficiency of data centers, should be a particular focus. Those data center operators who adeptly brace for AI’s demands will find themselves well-placed to leverage the forthcoming prospects accompanying its evolutionary journey and its widespread integration.

人工智能是一種顛覆性技術(shù)，它不僅有能力改變我們的職業(yè)生涯，還有可能改變我們的生存的基本構(gòu)架。數(shù)據(jù)中心運(yùn)營(yíng)商需要為此做好準(zhǔn)備。采取措施促進(jìn)數(shù)據(jù)速度的無(wú)縫轉(zhuǎn)變，并提高數(shù)據(jù)中心的能源效率，應(yīng)成為特別關(guān)注的重點(diǎn)。如果數(shù)據(jù)中心運(yùn)營(yíng)商能夠?yàn)槿斯ぶ悄艿男枨笞龊脺?zhǔn)備，那么他們就能很好地利用人工智能在發(fā)展歷程和廣泛融合的過(guò)程中進(jìn)行布局，讓自己在行業(yè)內(nèi)處于領(lǐng)先地位。

Trends in New AI Use Cases

新人工智能使用趨勢(shì)案例

While generative AI and related applications are rapidly evolving, several emerging areas already show significant potential in the near term. Some leading new AI applications to watch for include:

雖然生成式人工智能和相關(guān)應(yīng)用正在迅速發(fā)展，但一些新興領(lǐng)域已經(jīng)在近期顯示出的巨大潛力。值得關(guān)注的一些領(lǐng)先的新人工智能應(yīng)用包括:

Generative AI: Generative AI involves the creation of new data, such as images, text, music, or video, by AI models. This technology has applications in various fields, including content creation, design, gaming, and virtual reality. Generative adversarial networks (GANs), that are designed for iterative self-correction learning, have shown remarkable progress in generating realistic, high-quality content.

生成式人工智能（Generative AI）涉及由AI模型創(chuàng)造的新數(shù)據(jù)，如圖像、文本、音樂(lè)或視頻。這項(xiàng)技術(shù)在包括內(nèi)容創(chuàng)作、設(shè)計(jì)、游戲和虛擬現(xiàn)實(shí)等各個(gè)領(lǐng)域都有應(yīng)用。生成對(duì)抗網(wǎng)絡(luò)（GANs）是為迭代自我校正學(xué)習(xí)而設(shè)計(jì)的，已經(jīng)在生成逼真、高質(zhì)量?jī)?nèi)容方面取得了顯著進(jìn)展。
原創(chuàng)翻譯：龍騰網(wǎng) http://www.top-shui.cn 轉(zhuǎn)載請(qǐng)注明出處


AI in Natural Language Processing (NLP): NLP continues to evolve, with advancements in language understanding, sentiment analysis, and language generation. OpenAI's GPT models, for instance, have demonstrated impressive language generation capabilities. Future applications include more natural and conversational virtual assistants, improved language translation, and enhanced content creation.

自然語(yǔ)言處理 (NLP) 中的人工智能： NLP 不斷的發(fā)展下，在語(yǔ)言理解、情感分析和語(yǔ)言生成方面取得了進(jìn)步。例如，OpenAI 的 GPT 模型已經(jīng)展示了令人印象深刻的語(yǔ)言生成能力。未來(lái)的應(yīng)用包括更自然、更富對(duì)話性的虛擬助手、進(jìn)階的語(yǔ)言翻譯和內(nèi)容創(chuàng)作上的增強(qiáng)。

Edge AI: Edge computing combined with AI is gaining traction. By deploying AI algorithms and models directly on edge devices, such as smartphones, IoT devices, and autonomous vehicles, real-time decision-making and local data processing can be achieved. This enables faster response times, reduced latency, and improved privacy.

邊緣人工智能（Edge AI）：邊緣計(jì)算與人工智能的結(jié)合正在逐漸受到關(guān)注。通過(guò)在智能手機(jī)、物聯(lián)網(wǎng)設(shè)備和自動(dòng)駕駛汽車等邊緣設(shè)備上直接部署人工智能算法與模型，可以實(shí)現(xiàn)實(shí)時(shí)決策和本地?cái)?shù)據(jù)處理。這有助于加快響應(yīng)速度、減少延遲，并提高隱私性。

Explainable AI (XAI): Explainable AI focuses on making AI models and their decisions transparent and interpretable to humans. It aims to address the "black box" nature of deep learning models and provides insights into why certain decisions or predictions are made. XAI is crucial for building trust in AI systems, especially in domains like healthcare, finance, and law.

可解釋人工智能（XAI）：可解釋人工智能專注于使人工智能模型及其決策，對(duì)人類透明且易于理解。其目標(biāo)是解決深度學(xué)習(xí)模型的“黑匣子”特性，并提供關(guān)于為何做出某些決策或預(yù)測(cè)性的見解。XAI對(duì)于建立人們對(duì)人工智能系統(tǒng)的信任至關(guān)重要，特別是在醫(yī)療保健、金融和法律等領(lǐng)域。

AI in Robotics and Automation (XAI): The integration of AI with robotics is advancing automation capabilities across industries. Collaborative robots, or cobots, equipped with AI can work alongside humans in manufacturing and assembly tasks. AI also enables robots to learn and adapt to new environments, enhancing their autonomy and versatility.

機(jī)器人與自動(dòng)化領(lǐng)域的人工智能（XAI）： 人工智能與機(jī)器人技術(shù)的融合正在推進(jìn)各行業(yè)的自動(dòng)化能力。配備有人工智能的協(xié)作機(jī)器人（或 cobots）可與人類一起完成制造和裝配任務(wù)。人工智能還能讓機(jī)器人學(xué)習(xí)和適應(yīng)新環(huán)境，增強(qiáng)其自主性和多功能性。

AI for Cybersecurity: As cyber threats become more sophisticated, AI is being employed to strengthen cybersecurity measures. AI algorithms can detect anomalies in network traffic, identify patterns of malicious activity, and prevent cyber-attacks. AI-driven cybersecurity systems can respond and adapt to evolving threats in real-time, providing enhanced protection.

AI在網(wǎng)絡(luò)安全中的應(yīng)用：隨著網(wǎng)絡(luò)威脅變得更加復(fù)雜，人工智能被用于加強(qiáng)網(wǎng)絡(luò)安防。人工智能的算法可以檢測(cè)網(wǎng)絡(luò)流量中的異常，識(shí)別惡意活動(dòng)的模式，并預(yù)防網(wǎng)絡(luò)攻擊?；谌斯ぶ悄茯?qū)動(dòng)的網(wǎng)絡(luò)安全系統(tǒng)可以實(shí)時(shí)響應(yīng)和適應(yīng)不斷演變的威脅，為網(wǎng)絡(luò)安全提供了更加強(qiáng)大的保護(hù)。
原創(chuàng)翻譯：龍騰網(wǎng) http://www.top-shui.cn 轉(zhuǎn)載請(qǐng)注明出處


AI in Personalized Medicine: AI is revolutionizing healthcare by enabling personalized medicine approaches. Machine learning models can analyze large-scale patient data, genetic information, and medical records to identify patterns and correlations. This can aid in disease diagnosis, treatment sextion, and predicting patient outcomes, leading to more effective healthcare interventions.

個(gè)性化醫(yī)療中的人工智能： 通過(guò)實(shí)現(xiàn)個(gè)性化醫(yī)療方法，人工智能正在徹底改變醫(yī)療保健。機(jī)器學(xué)習(xí)模型可以分析大規(guī)模的患者數(shù)據(jù)、基因信息和醫(yī)療記錄，以確定模式和相關(guān)性。這有助于疾病診斷、治療選擇和預(yù)測(cè)患者預(yù)后，從而實(shí)現(xiàn)更有效的醫(yī)療干預(yù)。
原創(chuàng)翻譯：龍騰網(wǎng) http://www.top-shui.cn 轉(zhuǎn)載請(qǐng)注明出處


AI for Climate Change and Sustainability: AI is being explored to address pressing environmental challenges. It can help analyze climate data, optimize energy consumption, predict weather patterns, and develop sustainable solutions. AI-powered systems have the potential to optimize resource utilization, reduce emissions, and contribute to environmental conservation.

AI在應(yīng)對(duì)氣候變化和可持續(xù)發(fā)展方面的應(yīng)用：人工智能被用于解決緊迫的環(huán)境挑戰(zhàn)。它可以幫助分析氣候數(shù)據(jù)，優(yōu)化能源消耗，預(yù)測(cè)天氣模式，并開發(fā)可持續(xù)解決方案?；谌斯ぶ悄艿南到y(tǒng)有潛力優(yōu)化資源利用，減少排放，并促進(jìn)環(huán)境保護(hù)。
原創(chuàng)翻譯：龍騰網(wǎng) http://www.top-shui.cn 轉(zhuǎn)載請(qǐng)注明出處


AI in Natural Language Processing (NLP): NLP continues to evolve, with advancements in language understanding, sentiment analysis, and language generation. OpenAI's GPT models, for instance, have demonstrated impressive language generation capabilities. Future applications include more natural and conversational virtual assistants, improved language translation, and enhanced content creation.

AI在應(yīng)對(duì)氣候變化和可持續(xù)發(fā)展方面的應(yīng)用： 人們正在探索利用人工智能來(lái)應(yīng)對(duì)緊迫的環(huán)境挑戰(zhàn)。它可以幫助分析氣候數(shù)據(jù)、優(yōu)化能源消耗、預(yù)測(cè)天氣模式并制定可持續(xù)的解決方案?；谌斯ぶ悄艿南到y(tǒng)有潛力優(yōu)化資源利用，減少排放，并促進(jìn)環(huán)境保護(hù)。