Load Balancing & Web Dispatcher

SAP recommends server-based load balancing over client-based. What is the concrete operational payoff that drives that recommendation?

In client-based load balancing, how does a client end up talking to a specific application server, and by which two mechanisms is this realized?

Both the message server method and the SAP Web Dispatcher balance HTTP load. What is the fundamental behavioral difference, and what extra role does that let the Web Dispatcher fill?

Do you download or separately purchase the SAP Web Dispatcher, and which system landscapes can it load-balance?

Why is the message-server HTTP-redirect method not SAP's preferred load balancing approach? Name its practical drawbacks.

For a load balancer, why are stateful web applications the hard case rather than stateless ones?

AS Java performs load balancing at two levels. What distributes traffic at each level, and what is the difference between them?

The SAP Web Dispatcher never gets a static list of your Java instances. How does it learn the topology and capacity it needs to load-balance, and what minimal config does that require?

What kind of component is the SAP Web Dispatcher, and what is the first decision it makes on an inbound connection?

When the SAP Web Dispatcher weights how many requests each instance gets, what capacity metric does it use for an AS ABAP versus an AS Java instance?

The SAP Web Dispatcher keeps a stateful user on the same server process. Why does it use a different stickiness key for plain HTTP than for end-to-end SSL?

A team wants HTTP load balancing but does not want to install or configure any extra component. What is already available out of the box in SAP NetWeaver AS, and how does its flow work?