Dismiss
InnovationQ will be updated on Sunday, Oct. 22, from 10am ET - noon. You may experience brief service interruptions during that time.
Browse Prior Art Database

Video Stream Tagging with Data Center System Event Log Information

IP.com Disclosure Number: IPCOM000236303D
Publication Date: 2014-Apr-17
Document File: 2 page(s) / 23K

Publishing Venue

The IP.com Prior Art Database

Abstract

Disclosed is a system to tag the global surveillance video streams of a given area with event data from the servers in the associated data center. This creates a tagged video stream set that can be indexed, searched, and selectively saved for future reference without requiring the administrative staff to curate the data.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 51% of the total text.

Page 01 of 2

Video Stream Tagging with Data Center System Event Log Information

In data center server rooms, it is common to control physical access to machines for purposes of both security and availability. When improper access to machines occurs, video surveillance may be employed to help in forensic analysis of the access. Over a period, storage requirements to maintain an interruption-free video record of a data center become prohibitive. Additionally, automated video analysis may limit the extent to which a search may be narrowed down for a human to view for information. In current solutions, frame delta analysis is used to identify potentially interesting video sequences based upon movement of personnel within the field of view of the camera.

Video records are typically retained for a fixed period, based upon storage constraints, leaving the oldest historical records to be discarded to free-up capacity for new data. This leaves forensic analysis at a disadvantage for finding interesting information and having all historical information available for a thorough investigation. In some state-of-the-art implementations, video is only captured in response to a specific event, with the camera being integrated as a recipient of the alert for specific actions that are deemed interesting and within the field of view of the camera. For example, a rack-mounted camera module may capture a picture of a short video stream sequence based upon the triggering of a door sensor on a secure rack to ensure that the personnel accessing the door are recorded. The disadvantage of this technique is that the time leading up to the event is not captured unless the camera is continuously recording into a sufficient buffer to allow for that capture and the video stream produced is not annotated and therefore, not searchable.

The novel contribution is a system to tag the global surveillance video streams of the data center with event data from the servers in the data center. This creates a tagged video stream set that can be indexed, searched, and selectively saved for future reference without requiring the administrative staff to curate the data. The system first categorizes events as having potential for cause by physical interaction. Then, when relevant events are detected, the video stream(s) associated with the source location of the events are tagged accordingly. In extended embodiments, when the video stream storage is reclaimed, tagged segments can be selectively archived as part of the event record. This process can include digital analysis of the tagged stream to decide if there is any interesting video data to preserve in the event record.

This article describes the system with respect to a data center environment; however, this technique applies to any instrumented environment where devices emit alerts that are of interest when matched to a separate video surveillance stream. There are numerous examples, outside of the data center, of this abstract conc...