Cloud load balancing for low latency applications

低延迟应用程序的云负载平衡

• 批准号: 514068-2017
• 负责人: Feng, Chen
• 金额: $ 1.81万
• 依托单位: University of British Columbia
• 依托单位国家: 加拿大
• 项目类别: Engage Grants Program
• 财政年份: 2017
• 资助国家: 加拿大
• 起止时间: 2017-01-01 至 2018-12-31
• 项目状态: 已结题
• 来源: https://www.nserc-crsng.gc.ca/ase-oro/Details-Detailles_eng.asp?id=640414
• 关键词: Cloud; load; balancing; low; latency

Two Hat Security aims to develop next generation moderation tools to identify offensive images/videos,especially those containing child sexual abuse material (CSAM). If Two Hat is able to detect those contents assoon as they are uploaded to the cloud, it can help to stop the spread of CSAM online. However, this presents adifficult engineering challenge: After Two Hat develops a highly accurate tool to detect CSAM, how can itsupport low latency detection with ever increasing CSAM contents? The proposed project is to achievesub-second detection time in a cloud environment with a large number of servers. We will make use of a recentbreakthrough in cloud load balancing---the so-called Join-the-Idle-Queue (JIQ) strategy. First, we will conducta complete theoretical analysis for JIQ and in doing so obtain important engineering insights. Second, we willdevelop various JIQ algorithms based on our insights. Finally, we will work together with software engineers atTwo Hat to implement our JIQ algorithms on a cloud platform, such as Amazon EC2, Google Cloud, orMicrosoft Azure.

Two Hat Security旨在开发下一代审核工具来识别攻击性图像/视频，特别是那些包含儿童性虐待材料(CSAM)的图像/视频。如果Two Hat能够在这些内容上传到云中后立即检测到它们，则有助于阻止CSAM在线传播。然而，这带来了一个困难的工程挑战：在Two Hat开发了一个高精度的CSAM检测工具后，它如何在CSAM内容不断增加的情况下支持低延迟检测？提出的方案是在拥有大量服务器的云环境中实现亚秒检测时间。我们将利用云负载均衡方面的最新突破-所谓的加入空闲队列(JIQ)策略。首先，我们将为JIQ提供完整的理论分析，并在此过程中获得重要的工程见解。其次，我们将根据我们的见解开发各种JIQ算法。最后，我们将与Two Hat的软件工程师合作，在云平台上实现我们的JIQ算法，例如Amazon EC2、Google Cloud或Microsoft Azure。