I recently attended the "Power and Signal Integrity Characterization Techniques" class at IMS 2016.
The presenters were Heidi Barns: Keysite Technologies, Eric Bogatin: University of Colorado, Steven Sandler: PicoTest, and Jack Carrel: Xilinx.
Steven Sandler mentioned that one should never use ferrite beads for power filtering, rather RC networks or fast LDO's with low values of output capacitors.
I asked the question to the panel "Does this mean that all these years that all the applications notes recommended using ferrite beads to filter the power to sensitive circuits were wrong" There was a unanimous yes! They have been wrong all these years.
Considering this, why does your reference design for the AD9371 have 13 different power domains, all filtered with ferrite beads?
Previously I did a design with the TI Marconi chip where I removed all their similar power filtering with almost no measurable reduction in performance.
The large number of power domains makes PCB layout very difficult and costly.
Instead it would be nice if you could supply us with the current and noise requirements for each power domain.