EXPERT RESPONSE
Currently, there are two options for providing PoE. One is endspan (where the switch provides power) and the other is midspan where a device injects the power in the channel on the unused data pairs. Midspan is limited to 10/100. Although there are companies that say they have gigabit capable products, they are not standards based and with the standard roughly four years out, the requirements could change drastically.
There is a new working group in IEEE called Power over Ethernet plus. They'll write the project authorization request (PAR) and answer what is affectionately called the 5 critters (5 criteria to determine if it is worth the work to become a standard). If that is approved by the IEEE voting body, they then become a task force and the standard is typically four 4 years out. Many people will have an interest in the final verbiage of the standard, and this task group has lots of participants so anyone saying that they have a standard gigabit midspan solution right now is incorrect.
Currently, PoE is limited to low watt equipment. Deliverable power at the end device right now is 12.95W, although more power is provided at the source because some is lost through attenuation. The PoE plus group is working to double that amount of power to accommodate things like higher gain antennas. Two of the objectives are to increase the power and to have a gigabit midspan solution, neither of which are in the current 802.3af standard. Because endspan does not break into the channel, there is less degradation of the signal on the channel and therefore it can supply power (DC not AC so coupling of interfering signals is not an issue).
IEEE has sent liaison letters to the ISO and TIA to determine what the effect of the heat will be on data cables. Right now, that is a large unknown. Work is just beginning on this one, so it will be interesting to see where it all shakes out. There are also powered patch panels on the market which are 10/100 midspan devices. If a device loses power with these, a simple patch cable switch will not suffice, but rather a retermination would be required to move that channel to one with power. Not very smart for day two administrative costs.
However, the active electronics manufacturers have made it very economical and affordable to add power to switches. In my opinion this is the best way. You have one point to look for errors, you do not have to reterminate a connection of the power fails, you can add power where and when needed without retermination, and the new switches provide a power back off so that if a device does not need the whole 12.95W, a lower amount can be sent down the wire saving lots of day two costs for electricity. And it requires less rack space than many current options and as we know, floor space is at a premium.
Also, upgrades are typically chip replacements and require much less time. If you think about it, the active electronics manufacturers don't really want other manufacturer's gear in their closet if possible. Not only due to wanting to "own the closet" so to speak, but also because it decreases the complexity of troubleshooting and lowers the chance of interoperability problems.
Will it completely remove all power, I don't think so, at least not in the short term. Our devices still require more than that, but I did hear that they have come up with a man's electric razor that runs on PoE. The big advantage here is that network connectors are standard all over the world. Electrical connectors are not standardized - so it would be cool if you could run your whole briefcase via PoE. But it will be some time before all devices can run on that little power or the newer PoE plus is ratified. It certainly has its place, and is great when possible, but I believe that endspan (where the power is supplied by the network switch) will remain the victorious champ for a lot of reasons.
|
