专利快速检索-快速检索全球专利，免费商用专利数据库-IPRDB

1. 发明授权

US09584419B1 Systems and methods for detecting consistent fabric congestion in a datacenter based on historical traffic data 有权
标题翻译：基于历史流量数据检测数据中心中一致的布线拥塞的系统和方法
公开(公告)号：US09584419B1
公开(公告)日：2017-02-28
申请号：US14661901
申请日：2015-03-18
申请人： Google Inc.
发明人： Alexander Gilgur , Harpreet S. Chadha
IPC分类号： H04W4/00 , H04L12/801 , H04L12/26
CPC分类号： H04L47/11 , H04L43/0888
摘要： Detecting network congestion at a network device by collecting, for various time intervals, measurements of a quantifiable characteristic of traffic transiting the network device at a plurality of respective measurement times during each of the time intervals. Described systems and methods select two or more sets of measurements, each set corresponding to a respective time interval; determine, for each of the selected sets of measurements, respective upper and lower boundaries for the quantifiable characteristic during the respective time interval; and analyze trends for a time window inclusive of the time intervals corresponding to the selected sets of measurements, the trends including an upper boundary trend and a lower boundary trend. Congestion is then identified responsive to the analysis detecting a substantially constant upper boundary trend concurrent with a rising lower boundary trend.
摘要翻译：通过在各个时间间隔内收集在每个时间间隔期间的多个相应的测量时间内对网络设备进行的流量的可量化特征的测量来检测网络设备上的网络拥塞。描述的系统和方法选择两组或多组测量，每组测量对应于相应的时间间隔; 为每个所选择的测量集确定在相应的时间间隔期间可量化特征的相应上限和下限; 并且分析包括对应于所选择的测量集合的时间间隔的时间窗口的趋势，趋势包括上边界趋势和下边界趋势。然后响应于分析检测到拥塞，该分析检测与上升的下边界趋势同时发生的基本上恒定的上边界趋势。

IPRDB

热门服务

关于我们

友情链接

联系方式