Check your VMS prior to installing the cameras
Check your network load prior to installing the cameras
Check your Operation Center prior to installing the cameras
Audit and benchmark of large scale ip video systems
A video surveillance system is composed of three major layers : from the edge to the core, the sensors (cameras), the network and the computer infrastructures including video management systems and operator workstations.
During the design phase, assessing the servers capacity is a key but delicate step that only relies on constructors and software vendors abacus and computations. While these data allow a straightforward computation of the required bandwidth or storage capacity required for the system (related to the number of cameras, number of frames per second, number of days video must be kept) they fail giving an accurate idea of the CPU power necessary for the operator viewing PCs.
VMS software vendors cannot either give any precise information about CPU power or graphic card for the operators workstations since it depends of the use of the VMS (Video Management System), of the number of screens per station, of the number of concurrent camera feeds per screen.
During installation and commissioning phases problems happen, that should have been resolved during design phase.
Beside, launching a test phase of the operation control room (urban operation center for instance), is strictly conditionned by the installation and commissioning of the cameras and the network. It is not possible to test VMS software and network video recorders in real scale prior to that.
An elegant solution to this set of project scheduling problems is carried out by the Xstream Generator software from DOO Technologies company. It allows to record video samples from a vast panel of Onvif compliant cameras and to use those footages to emulate cameras by producing recorded video endlessly.
Very effective, Xstream Generator needs few CPU and RAM per virtual camera on the Windows PC running it. It becomes possible to emulate more than 1000 cameras on a high power PC, with network bandwidth as sole real limitation.
This post is also available in: French