- Use a subnet broadcast in the layer three destination address to wake up a client. An example of which would be sending the packet to 192.168.3.255 to get to a host on the 192.168.3.0 subnet. In the layer 2 (AKA "ethernet header") address portion of the data frame, the destination address would be the MAC address of the intended target system. If subnet broadcasts are not disallowed on the routers access list, that technique alone gets the WOL traffic through the WAN's routers to the desired WOL client.
- Send an IP layer 3 broadcast (255.255.255.255) with the target WOL systems MAC address in the Layer 2 frame destination address header; it is not going to work across a WAN segment because most WAN admins (understandably) will never consider allowing that type of traffic to go across the enterprise routers.
The mention of the MAC address tables on the document can be a little misleading as the MAC address of a target host is only kept in a table on the routers that have an interface on the client's home subnet (e.g. the target host's default gateway router). The implication of this is that MAC addresses are not only subject to being dropped from a routers MAC address table, as the document states, they are also not routable (that is what layer 3 protocols [IP, IPX/SPX, and so on] are for <g>). In the low power state, the client's NIC has no IP address to send to, so the only way to address the routing issue would be via some form of broadcast.
WOL Proxies do not replicate any packets, it sends new ones on its local segment. Specifically select clients to be multicast and WOL proxies.
When the server tries to send Wake on LAN packets it will check the database for any clients that are Wake on LAN Proxies. The server will send the Wake on LAN packets and tell each Wake on LAN Proxy to send the same packets. It tells every proxy to send every packet every time because there is no intelligence included to determine which WOL proxy should send out which packets.
The UDP WOL port is the same port the client uses to connect to the Deployment Solution server.