By itself, HSRP/GLBP does not provide higher availability of redundant WAN links (the key feature request in the original poster's question). WAN link load balancing is a side effect of HSRP/GLBP only in a very limited scenario where the source of traffic is from multiple local systems on the same LAN segment as both gateway routers and the choice of default gateway has the side effect of selecting one WAN link over another. There is no scenario where HSRP/GLBP provides load balancing for inbound traffic from the WAN.
Depending upon the specific connectivity scenario, the appropriate solution for higher availability could include keepalives, frame-relay end-to-end keepalives, EIGRP, OSPF, RIP, RIPv2, floating static routes, BGP, policy routing on reachability, etc., and appropriate solutions for load balancing could include CEF per packet, CEF per flow, MLPPP, BGP AS path padding, creative IP subsetting, EIGRP variance, policy routing, etc.
You are correct that HSRP/GLBP may have a place in a design with WAN links (and, indeed, I give an example of exactly that usage in my book High Availability Networking with Cisco). But to call it a solution for the problem asked is giving it too much credit, and could contribute to the common misconception that HSRP/GLBP is a substitute for a routing protocol rather than a mechanism to get around the shortcomings of IP's default gateway limitations on a LAN with multiple routers present.
Vincent C Jones
www.networkingunlimited.com
