In my last post, I began enumerating the top 10 issues and trends that will affect IP surveillance in the next few years. Part 1 discussed gauging the impact IP-based cameras will have on the enterprise network in terms of bandwidth, the importance of storage tiering and an explicit policy regarding how long digital imagery is stored, and some of the security issues that arise when you move from analog tape to digital storage and transmission.
Now, let’s take a look at three more considerations when implementing IP surveillance, with a decidedly hardware-focused slant.
1. Switching
Now that you’re no longer dealing with analogue signals, your digital video has to climb aboard the enterprise network, and that’s where switches come in. Your bandwidth needs will vary depending on a number of factors—resolution, frames per second, compression algorithm—but with switches available from one gigabit per second (Gbps) to 100 Gbps, with stops in between at 10 and 40 Gbps, switching speed will probably be a minimal issue.
Depending on the environment, you may also want to consider ruggedized switches that are vibration, weather, and dirt resistant. If you have no choice but to mount a switch in an outdoor environment, this is probably the way to go.
There’s a move underway in the networking industry (in some circles, anyway) to push more intelligence out to the network edge. Intelligent switches allow you to perform sophisticated processing of the data your cameras are collecting before passing it through the network, or having that data control automated devices without the north-south trip to the data centre.
Is  cabling the cameras themselves problematic? Consider wireless access points (WAP) to connect your camera to the network (although, technically, that goes beyond the switching layer and into the routing layer of the network model).
2. Power over Ethernet (PoE)
Just because you can get an ethernet cable to that camera in the far corner of the parking lot doesn’t mean there is going to be a convenient source of 120-volt power, and batteries can fail and have to be replaced just when you need them.
PoE-enabled cameras and switches solve this problem by transmitting both data and electrical power (up to 50 watts) over the same cable, since two of the four pairs of wires in an ethernet cable aren’t required for data transmission. Depending on compatibility issues, a splitter may be required at the device end to tap the power and data connections.
3. PC servers
In an earlier post, I wrote that some elements of camera evolution are actually outpacing Moore’s Law. However, the IP-based camera is not running away from personal computer-based servers in this race anytime soon. Multicore processors, huge volumes of high-speed DRAM, solid state hard drives and dwindling storage costs are all ensuring the PC stays ahead of the game.
In general, higher frame rate and resolution put more demand on storage, while analytics processing of the data requires more RAM and processing power. If you’re planning on harnessing the power of analytics, it makes sense to separate the management of the cameras and the analytical load, putting the latter on a separate machine, or even leveraging cloud computing—but that’s a discussion for another day.
