Can you be a bit more specific? A google search for 'fault tolerance' returns about 843,000 hits.
And I suspect there is no single 'most effective and reliable' technology/design/product/whatever with regard to fault tolerance. It's all about trade-offs. Do you go with a single core switch with redundant supervisors & power supplies or two core switches, each with a single sup & ps? Do you design your network with layer 2 loops [ie. redundant links & rely on spanning tree to keep the network loop free] or do you go with no layer 2 redundancy and rely on the layer 3 routing protocol to achieve fault tolerance? A combination of both? And at what point do you decide when there's enough fault tolerance? Personally, I like the 'no single point of failure' guideline but YMMV
Fault tolerance is one of those subjects where there is no single correct answer - except maybe for "it depends"
regards,
Lee