发明名称 Acquisition and kernel memory storage of I/O metrics
摘要 An improved technique for storing I/O metrics includes assigning metric values to data buckets held in kernel memory. Each data bucket covers a particular range of values of a respective metric and is configured as a counter, whose count is incremented each time the multipathing driver obtains or computes a metric value that falls within the range of the data bucket. Bucket counts can be read by an external program to obtain aggregate information about I/O metrics over time. The aggregate information can be fed back to the multipathing driver to enable improved selections of paths for conveying data to and from a storage array.
申请公布号 US8751757(B1) 申请公布日期 2014.06.10
申请号 US201113341348 申请日期 2011.12.30
申请人 EMC Corporation 发明人 Gaikwad Deepak M.;Pellowski Robert J.;Epstein Edith;Trivedi Hitesh;Raizen Helen S.
分类号 G06F12/00 主分类号 G06F12/00
代理机构 代理人
主权项 1. A method operable on a computing device for obtaining performance information relating to data storage, comprising: providing, in a software driver running within a kernel of the computing device, multiple data buckets for storing information pertaining to a metric relating to data storage, each data bucket covering a different range of values of the metric; monitoring I/O operations attempted between the computing device and at least one logical storage unit (LUN) to produce multiple values of the metric; assigning each value of the metric to one of the data buckets for the metric; counting the number of values of the metric assigned to each data bucket to produce multiple data bucket totals; and discarding the counted values of the metric, wherein each value of the metric is related to a respective I/O request, and wherein monitoring I/O operations attempted between the computing device and at least one logical storage unit to produce multiple values of the metric includes, for at least one of the multiple values: detecting an arrival to the software driver of an I/O request and associating a first timestamp therewith;detecting a passage of the I/O request from the software driver and recording a second timestamp associated therewith;detecting a return to the software driver of a returning I/O response and associating a third timestamp therewith; andcomputing a latency value based on any two of the first through third timestamps.
地址 Hopkinton MA US